1
Stanford University John A. and Cynthia Fry Gunn Building 366 Galvez Street | Stanford, CA | 94305-6015
Working Paper No. 525
Tracing Greenhouse Gas Emissions
in Global Value Chains
By
Bo Meng Glen P. Peters
Zhi Wang
April 2015
2
Tracing Greenhouse Gas Emissions in Global Value Chains*
Bo MENG, Glen P. PETERS, and Zhi WANG
Abstract
This paper integrates two lines of research into a unified conceptual framework: trade in global
value chains and embodied emissions. This allows both value added and emissions to be
systematically traced at the country, sector, and bilateral levels through various routes in global
production networks. By combining value-added and emissions accounting in a consistent way,
the potential environmental cost (amount of emissions per unit of value added) along global
value chains can be estimated from different perspectives (production, consumption, and trade).
Using this unified accounting method, we trace value-added and CO2 emissions in global
production and trade networks among 41 economies in 35 sectors from 1995 to 2009 based on
the World Input–Output Database, and show how they improve our understanding on the impact
of cross-border production sharing on the environment.
Keywords: trade in value-added; embodied emissions; carbon intensity; global value chains and the environment JEL Codes: E01, E16, F1, F14, F18, Q5, Q54, Q56
* Meng: Institute of Developing Economies, Chiba, Japan ([email protected]); Peters: Center for International Climate and Environmental Research, Oslo, Norway (glen.peters@ cicero.oslo.no); and Wang: US International Trade Commission, Washington, D.C. ([email protected]). The views in the paper are those of the authors and may not reflect the views of the USITC and its Commissioners or any other organization that the authors are affiliated with. We thank Jinjun Xue (Nagoya University, Japan) and Kunfu Zhu (Chinese Academy of Sciences) for discussions related to this work and IDE-JETRO’s peer reviewer, Dr. Masami Ishida as well as the participants of the 2014 Beijing Humboldt Forum for their comments on an earlier version of the draft. We are grateful for the financial support from CIDEG, Tsinghua University. Zhi Wang also acknowledges the research support provided by Stanford Center for International Development.
3
Tracing Greenhouse Gas Emissions in Global Value Chains
1. Introduction
The rise of global value chains (GVCs) during the last two decades has significantly changed the
nature and structure of international trade, with many new implications for policy (Baldwin,
2012; Timmer et al. 2013). Studies on GVCs have covered a variety of topics such as vertical
specialization (Hummel et al. 2001), trade in tasks (Grossman and Rossi-Hansberg 2008;
Baldwin and Robert-Nicoud 2014), magnification of trade cost from multi-stage production (Yi
2010), value chain organization (Antras and Chor 2013) as well as the measurement of the
creation and distribution of employment and income in GVCs (OECD et al. 2013; Timmer et al.
2014b; Ferrarini and Hummels 2014).
In recent years, however, many scholars have turned their attention to the interaction of
GVCs and environmental policies (Hoekstra and Wiedmann 2014). A large body of literature has
developed to assess “consumption-based accounting” of historical emissions (Tukker and
Dietzenbacher 2013). This literature adjusts the standard territorial-based emission accounts by
removing the emissions associated with the production of exports and adding the emissions
associated with the production of imports (Peters and Hertwich 2008). Most early studies
focused on climate policy. It has been found that international trade accounts for one-quarter of
global carbon emissions, but the contributions of exports to a country’s territorial emissions
(median 29%, range 8–64%, year 2007) and imports to a country’s consumption-based emissions
(median 49%, range 6–196%, year 2007) are significant (Andrew and Peters 2013). Developed
nations collectively have higher consumption-based emissions than territory-based emissions,
meaning that they are net importers of emissions and thereby benefit from environmentally
intensive production abroad (Davis and Caldeira 2010; Peters et al. 2011; Arto and
Dietzenbacher 2014).These effects are growing over time, and the net transfer of emissions
(production minus consumption) via international trade from developing countries to developed
countries increased from 0.4 Gt CO2 in 1990 to 1.6 Gt CO2 in 2008, which exceeds the emissions
reductions obtained within the Kyoto Protocol (Peters et al. 2011). The same conclusions have
been reached for many environmental issues, such as energy (Davis et al. 2011), air pollution
(Lin et al. 2014), material use (Wiedmann et al. 2013), land use (Weinzettel et al. 2013), biomass
4
(Peters et al. 2012), water (Hoekstra and Mekonnen 2012), and biodiversity (Lenzen et al. 2012).
For example, Lin et al. (PNAS, 2014) shows that 12-24% of sulfate concentrations over the
western United States on a daily basis is due to the export-related Chinese pollution, and Lenzen
et al. (Nature, 2012) discovered that about 30% of global species threats are due to international
trade.
The research on consumption-based accounting of environmental impacts has
considerable methodological and conceptual overlap with the work on trade in value added
(Johnson and Noguera, 2012, Koopman et al. 2014, Timmer et al. 2014b), but so far there has
been very little attempt to formally link these two independent lines of research. This is the
objective of this paper.
Better understanding the relationship between emissions and GVCs requires a consistent
and well-defined accounting system, which can provide proper measurements to trace value
added and the amount of emissions in each stage of production and trade from different
perspectives along the GVCs consistently and systemically.
In building such a unified accounting framework, existing efforts toward the
measurement of embodied emissions in trade, based on multi-regional input–output (MRIO)
models, provide a good starting point (e.g., Peters 2008; Peters and Hertwich 2008; Hertwich and
Peters 2009; Kanemoto et al. 2012; Meng et al. 2013). These efforts have significantly enhanced
our understanding of embodied emissions in trade, and provide complete account of embodied
emissions in global supply chains at country aggregates. However, less attention has been paid to
the difficulties to associate embodied emission with gross bilateral trade flows, especially at the
sector/product level (Atkinson et al., 2011), thus limits its policy relevance such as border carbon
tax design (Atkinson, 2013 ).
By integrating recent international trade literature on gross trade accounting and
environment economics literature on embodied emission trade and carbon footprint, this paper
makes the following new contributions:
First, we generalize existing measures of embodied emissions and consistently define
trade-related embodied emission measures at country, industry, bilateral and product levels in
5
precise mathematical terms. We also define trade in emission measure that is fully consistent
with gross bilateral trade flows, overcoming incompleteness of existing measures1.
Second, by integrating with gross trade accounting methods in recent international
economics literature, we are able to measure trade in value-added and trade in emissions at
country, bilateral, and sector/product levels in one unified accounting framework. Such a
framework is not only able to measure value-added and emissions generated from each
production stage (slice the value chain), but can also identify the special trade routes by which
value-added and emissions are created, transferred, and consumed. By combining value-added
and emissions accounting in a consistent way, the potential environmental cost along GVCs can
also be estimated (e.g. emissions with per unit of value-added created) from different
perspectives (production, consumption and trade).
Third, we demonstrated that the distinction between the forward and backward industrial-
linkage is the key to properly measure embodied emissions at disaggregate level. Building on
decomposition techniques originally developed by Leontief (1936), we show that using the
forward industrial-linkage-based decomposition, the total emissions from a country/industry can
be traced according to where and by which downstream GVC routes their associated gross output
are used. Using the backward industrial-linkage-based decomposition, we show that the total
emissions from all upstream production stages of a final good or service in a global value chain
can be fully identified. Both decomposition methods produce the same total emission estimates
for a country at the aggregate level, but they differ at the sector level due to differences in
measuring indirect emissions generated from production sharing arrangements.
Fourth, We follow the idea presented in the recent innovative work of Koopman et al.
(2014), and Wang et al. (2013), in which they decompose all bilateral intermediate trade flows
according to their final destination and express gross intermediate trade flows as destination 1 The existence of both Bilateral Trade Input-Output (BTIO) and Multi-Regional Input-output (MRIO) based measures in the large body of embodied emissions literature is due to two reasons: 1) when MRIO table is not available, using national IO table and international trade statistics, embodied emissions in bilateral trade can still be estimated. However, biases may occur since trade in intermediate exports is treated as exogenous variable in a BTIO model. 2) Using MRIO can remove such biases but once intermediate trade is treated as endogenous variable, the difficulty will come from how to properly allocate embodied emissions in gross intermediate trade flows. This remains unsolved in the existing literature until this paper. In this sense, we unified the two analytical frameworks into one system and enabled it to provide all emission measures derived from both MRIO and BITO in the existing literature.
6
countries’ final demands. Applying this technique to measure global emissions in gross exports,
we present a bridge to consistently link production-based and consumption-based accounts of
emissions at the regional, sectoral, and bilateral levels. We further decompose emissions
generated from the production of a country’s gross exports into eight different routes along
GVCs as well as their relative economic benefit/environment cost ratio first time in the literature.
We also separate emissions generated from production of a country’s GDP into international
trade related and unrelated portions, thus clearly distinguish emissions of self-responsibility
(emissions from production satisfies domestic final demands without through international trade)
and shared responsibility (emission from production satisfies domestic final demands through
international trade) between producers and consumers located in different territories.
Finally, we report a number of applications based on the World Input–Output Database
(WIOD2) to illustrate the potential of this new integrated accounting frameworks to deepen our
understanding of the impact of global value chains on the environment. For example, by clearly
distinguishing emissions generated from different GVC production routes, we find that
environmental cost for generating one unit of GDP only through domestic routes is lower than
that created through international trade for most G-20 countries in recent decades. The main
driver is the high-carbon-intensity trade in intermediates, which has grown rapidly during the
period we have data (1995-2009). More importantly, previous literatures emphasis emission
transfers between developed and developing countries, while the ability to decompose both
value-added and emission production and absorption by GVC routes enable us find such transfer
also happens among developing countries, and is increasingly becoming the major source of
emission transfer in the global production system, especially between China and other non-
Annex B countries (developing economies). Their share in total global trade related emissions
had increased dramatically from just 5% of in 1995 to nearly 20% in 2009. We also provide a
number of interesting figures that clearly show a country’s pattern and level of emissions is
crucially subject to its position and the extent of its participation, directly or indirectly, in GVCs
through international trade.
2 For detailed information, see Timmer et al. (2014a).
7
This paper is organized as follows: Section 2 presents the integrated accounting
framework and defines various embodied emission measures. Section 3 presents a number of
illustrative applications for tracing CO2 emissions in GVCs. Section 4 concludes.
2. Concepts and Methodology
2.1 Embodied emissions through forward and backward industrial linkage
The methods used to estimate embodied emissions3 are rooted in the work of Leontief (1936).
Leontief demonstrated that the complex linkages among different industries across countries can
be expressed as various inter-industry, cross-country transactions organized into chessboard-type
matrices, known as IO tables. Each column in the table represents the required inputs from other
industries (including imports and direct value added) to produce the given amount of the product
represented by that column. After normalization, the technical coefficient table represents the
amount and type of intermediate inputs needed in the production of one unit of gross output.
Using these coefficients, the gross output in all stages of production that is needed to produce
one unit of final products can be estimated via the Leontief inverse. When the output associated
with a particular level of final demand are known, the total emissions throughout the (global)
economy can be estimated by multiplying these output flows with the emission-intensity
coefficient (amount of emissions per unit of gross output) in each country/industry.
To illustrate how the classic Leontief method works, let us assume a two-country (home
and foreign) world, in which each country produces tradable products in N differentiated
industries. Products in each sector can be consumed directly or used as intermediate inputs, and
each country exports both intermediate and final products. All gross output produced by country
s must be used as either an intermediate or a final product at home or abroad, that is
Exports
srrsr
Domestic
ssssss YXAYXAX +++= r, s = 1,2 (1)
where Xs is the N×1 gross output vector of country s, Ysr is the N×1 final demand vector that
gives demand in country r for final goods produced in s, and Asr is the N×N IO input coefficient
3A clarification is needed on what is meant by “embodied”. The emissions embodied in gross output/final goods or exports/imports can be defined as the emissions that occur in the production of a product. The emissions are not actually a physical part of the product, but rather, are emitted in the production of the product.
8
matrix, giving intermediate use in r of goods produced in s. The superscripts in Asr and Ysr mean
that s is the producing country and r is the destination country. In (1), AssXs+Yss is domestic use
of products, while AsrXr+Ysr is exports to foreign countries, these in turn can be split into
intermediate use AssXs+AsrXr and final consumption Yss+Ysr. The two-country production and
trade system can be written as a multi-regional IO (MRIO) model in block matrix notations
++
+
=
rrrs
srss
r
s
rrrs
srss
r
s
YYYY
XX
AAAA
XX , (2)
which shows a clear distinction between intermediate use (AX) and final consumption (Y). The
intermediate use can be either at domestic market (diagonals) or exported to/imported from (off-
diagonals) foreign countries, and likewise for the final consumption. In this model, the final
consumption is exogenous, while intermediate use is endogenous. After rearranging terms, we
have
, (3)
where Bsr denotes an N×N block matrix, commonly known as the Leontief inverse, which is the
total requirement matrix that gives the amount of gross output in producing country s required
for a one-unit increase in final demand in country r. The diagonal terms Bss differ from the
“local” Leontief inverse 1)( −−= ssss AIL due to the inclusion of off-diagonal terms via the inverse
operation. Ys is an N×1 vector that gives global use of final products from country s, including
domestic final products sales Yss and final products exports Ysr.
For our later sector level analysis, it is worthwhile to break Equations (2) and (3) into
sectoral details. For N=2, this can be re-written by element as follows:
=
++
−−−−
=
−
r
s
rrrs
srss
rrrs
srss
rrrs
srss
r
s
YY
BBBB
YYYY
AIAAAI
XX
1
Domestic IO
Coefficients
9
++++
+
=
rrrs
rrrs
srss
srss
r
r
s
s
rrrrrsrs
rrrrrsrs
srsrssss
srsrssss
r
r
s
s
yyyyyyyy
xxxx
aaaaaaaaaaaaaaaa
xxxx
22
11
22
11
2
1
2
1
22212221
12111211
22212221
12111211
2
1
2
1
(2a)4
=
++++
−−−−−−−−−−−−−−−−
=
−
r
r
s
s
rrrrrsrs
rrrrrsrs
srsrssss
srsrssss
rrrs
rrrs
srss
srss
rrrrrsrs
rrrrrsrs
srsrssss
srsrssss
r
r
s
s
yyyy
bbbbbbbbbbbbbbbb
yyyyyyyy
aaaaaaaaaaaaaaaa
xxxx
2
1
2
1
22212221
12111211
22212221
12111211
22
11
22
11
1
22212221
12111211
22212221
12111211
2
1
2
1
11
11
, (3a)
where each element above is now a scalar: sjx is the gross output of sector j in country s; sr
iy
represents final goods produced by sector i in country s for consumption in country r (i,j = 1,2); srija is the direct IO coefficient that measures the intermediate inputs produced in sector i of
country s that are used in the production of one unit of gross output in sector j of country r, and srijb is the total requirement coefficient that gives the total amount of the gross output of sector i
in country s needed to produce an extra unit of the sector j’s final product in country r. Other
coefficients have similar economic interpretations.
Define the direct emission intensity as cj
cj
cj xpf ≡ for c = s,r, j=1,2, then the estimation and
decomposition of the country- and sector-level production of emissions can be expressed as
=∧
r
r
s
s
rrrrrsrs
rrrrrsrs
srsrssss
srsrssss
r
r
s
s
yy
yy
bbbbbbbbbbbbbbbb
ff
ff
YBF
2
1
2
1
22212221
12111211
22212221
12111211
2
1
2
1
000000000000
000000000000
ˆ
4 The elements in the diagonal block of the A matrix are domestic input-output coefficients, while elements in the off-diagonal block are import input –output coefficients. The Y matrix is similar.
Import IO
Coefficients
10
=
rrrrrrrrsrsrsrsr
rrrrrrrrsrsrsrsr
rsrsrsrsssssssss
rsrsrsrsssssssss
ybfybfybfybfybfybfybfybfybfybfybfybfybfybfybfybf
2222121222221212
2121111121211111
2222121222221212
2121111121211111
(5)
This matrix gives estimates of the sector and country sources of emissions in each
country’s final goods production. Each element in the matrix represents emissions from a source
industry of a source country directly or indirectly generated in the production of final products
(consumed in both the domestic and foreign markets) in the source country. Looking at the
matrix along the rows yields the distribution of emissions created from one country/sector across
all countries/sectors. For example, the first element of the first row, , is the
emissions created by sector 1 in country s to produce its final goods for both domestic sales and
exports. The second element, , is the emissions generated by sector 1 in country s
to produce intermediate input used by sector 2 in country s to produce its final products. The
third and fourth elements, and , are, respectively, emissions from
sector 1 in country s generated in the production of intermediate inputs used by the 1st and 2nd
sectors in country r to produce country r’s final products. Therefore, summing up the first row of
the matrix, we obtain the total emissions generated from sector 1 in country s. This can be
expressed mathematically as
[ ] [ ]rrsrsrrsrssrsssssssrssrsrssrsssssssssss
rsrrsrssssssssss
ybfybfybfybfybfybfybfybfybybybybfxfp
212111112121r
11112121111121211111
2121112121111111 )(
+++++++=
+++== (6)
which distributes the total emissions produced in a country/industry according to where its total
gross output are finally absorbed. The value of sjp is consistent with the production-based
National Emission Inventory (NEI) according to the economic activities of residential
institutions as defined by the System of National Accounts (SNA), similar to GDP by-industry
statistics (de Haan and Keuning 1996, 2001; Pedersen and de Haan 2006)5.
5For the difference between the production-based NEI estimates from the MRIO table and the UNFCCC NEI, see
)( 11111srsssss yybf +
)( 22121srsssss yybf +
)( 11111rrrssrs yybf + )( 22121
rrrssrs yybf +
11
Looking at the YBF ˆ∧
matrix down a column yields emissions estimates from all
countries/sectors across the world for the production of final products in a particular
country/sector. For example, the second element in the first column, )( 11212srsssrs yybf + , is the
amount of emissions generated in sector 2 of country s to produce intermediate inputs used by
sector 1 in country s to produce final products, and the third and fourth elements,
)( 11111srssrsr yybf + and )( 11212
srssrsr yybf + , respectively, are emissions generated in sectors 1 and 2 of
(foreign) country r to produce intermediate inputs used by sector 1 in country s in the production
of final products.
Adding up all elements in the first column gives the global emissions generated by the
production of final products in sector 1 of country s, denoted as )( 1syp , i.e
srsrrsrsssssss ybfbfbfbfyp 12121112121111 )()( +++= , (7)
It traces total emissions generated by the production of a final product in a particular
country/industry according to where the needed intermediate inputs are produced along each
stage (represented by different industries located in different countries) of the global production
chain. This is the global “carbon footprint” of the consumption of sector 1’s products from
country s. The last two terms represent imported emissions.
In summary, the sum of the YBF ˆ∧ matrix along a row represents the production-based
emissions and shows how each country’s emissions in a particular sector are distributed to final
consumption (across columns) of all downstream countries/sectors (including itself), thus
decomposes each country’s total emissions by industry according to where the final consumption
is made. It traces forward industrial linkages (downstream) from an emitter’s perspective. The
sum of the YBF ˆ∧ matrix along a column accounts for all upstream countries/sectors’ emissions
to the production of a specific country/sector’s final products (carbon footprint); it traces
backward industrial linkages across upstream countries/industries (as different stages of
production) from a user perspective, thus decomposes the total global emissions from the
Peters (2008).
12
production of a country/sector’s final goods and services according to where each of the needed
intermediate inputs is produced.
As an example, in the chemical sector, the producer’s perspective includes the emissions
created by the production of chemicals that are embodied in the final goods exports of chemical
products themselves (direct domestic emissions exports), as well as in the final exports of metal
products, computers, consumer appliances, and machineries that use chemicals as inputs (indirect
domestic emissions exports). Such a forward linkage perspective is consistent with the literature
on the emissions content of trade. On the other hand, decomposition from a user perspective
includes all upstream sectors/countries’ contributions to emissions in a specific sector/country’s
final goods exports. For instance, in the automobile industry, it includes emissions generated in
the automobile production itself as well as emissions embodied in inputs from all other upstream
sectors/countries (such as rubber from country A, glass from country B, steel from country C,
design and testing from the home country) used to produce an automobile for exports by the
home country. Such a backward industrial-linkage-based perspective aligns well with case
studies of emissions by a specific final product in the literature.
Each of these two different ways to decompose global total emissions has its own
interpretations and thus different roles in environmental policy analysis. The decomposition of
emissions by producing industry can address questions such as “who generates the emissions for
whose consumption?” thus providing a starting point for the discussion of shared responsibility
between producer and consumer at the industry level; while the decomposition of total emissions
generated to produce a final product is able to answer questions such as “what is the global
emissions level and what is the emission source (country/industry) structure required to produce
a car in Germany compared to that for China?” and can attribute the total emissions for a final
product to each stage of production in the global supply chain, thus providing facts that improve
our understanding of the common but differentiated responsibilities among different production
stages along each global supply chain.
With a clear understanding of how total national emissions by industry and total global
emissions by the production of final goods and services at the country-sector level can be
correctly estimated and decomposed by the standard Leontief method (equation (5) or the YBF ˆ∧
13
matrix), we formally specify the decomposition methods used in this paper and their relations to
other IO model based methods widely used in the literature.
2.2 Downstream decomposition: Decompose emissions generated from a country/industry based on forward industrial linkage
Extending equation (2) to a G country setting, the gross output production and use balance, or the
row balance condition of a MRIO table becomes
*ssssssG
rs
srsssssG
rs
srssG
rs
rsrssss EYXAEYXAYYXAXAX ++=++=+++= ∑∑∑≠≠≠
(8)
where ∑≠
=G
rs
srs EE * is the total gross export of country s. Rearranging (8) gives
*11 )()( ssssssss EAIYAIX −− −+−= (9)
With a further decomposition of the gross exports into exports of intermediate/final products and
their final destination of absorption, it can be shown that
∑ ∑∑∑∑∑
∑∑
≠ ≠≠≠
−
≠≠
≠≠
−−
+++−+=
+−=−
G
sr
G
rst
rtsrG
sr
rrsrG
sr
srssssssrsG
sr
srrsG
sr
sr
G
sr
rsrG
sr
srsssss
YBYBYBYAIABYB
XAYAIEAI
,
1
1*1
)(
)()()( (10)6
Inserting (10) into (9) and pre-multiplying the direct emission intensity diagonal matrix
F∧
, we obtain an equation that decomposes total emissions by industry into different
components.
)5()4()3()2()1(,
∑ ∑∑∑∑ ∑≠ ≠≠≠≠
∧∧∧∧∧∧
++++==G
sr
G
rst
rtsrG
sr
rrsrG
sr
srssG
sr
tsG
t
rtsrssssssss YBFYBFYBFYBALFYLFXFP ssssss
(11)7
Here, 1)( −−= ssss AIL is the local Leontief inverse. 6A detailed mathematical proof of equation (10) is provided in Appendix A.1. 7 The second term (2) on the right side in equation (11) equals to the sum of the first two terms on the right side in equation (10) (for detailed proof, see the appendix in Wang et al. 2013)
14
There are five terms in equation (11), each of which represents emissions generated by
the industry in its production to satisfy different segments of the global market. All the emissions
that occur in region s are a result of various elements of production.
• The first term: domestically produced and consumed final goods and services
(LssYss).
• The second term: domestically produced intermediate goods exports ( tsG
t
rtsrss YBAL ∑ )
which are used by other countries to produce either intermediate or final goods and
services shipped back to the source country as imports and consumed there. 8
• The third term: domestically produced final goods and service exports that are
consumed by all of its trading partners ( srssYB ).
• The fourth term: domestically produced intermediate goods and services exported to
country r for the production of final products consumed in country r ( rrsrYB )
• The fifth term: domestically produced intermediate goods exports to other countries
producing their final goods and service exports to third countries rtsrYB ).
Note the summation in the last three terms indicates that these emissions generated by
export production can be further split into each trading partner’s market. The sum of the last
three terms gives the amount of emissions exports, and the sum of the last four terms in each
bilateral route is the “Emissions Embodied in Bilateral Trade” (EEBT). Both measures are
frequently used in the literature on embodied emissions in trade, which we will discuss in detail
later in this paper. The disaggregated accounting for total emissions by industry based on
forward industrial linkage (downstream decomposition) made by equation (11) is also
diagrammed in Figure 1. The number in the lowest level box corresponds to the terms in
equation (11).
Figure 1 GHG emissions production, by sources of final demand – Forward industrial-linkage-based decomposition
8This indicates the second term in (11) can be further split according to a country’s final goods and intermediate goods imports and each particular trading partner that the imports come from.
A country/sector’s total
GHG emissions
Generated in production of domestic consumed
final goods and services
Generated in production of final goods and services
consumed abroad
15
2.3 Upstream decomposition: Decompose emissions from final goods and services by production stages in a global supply chain based on backward industrial linkage
In the following we estimate the total emissions generated by a final product along the global
supply chain identified by the last stage of production: a particular industry i located in a specific
country s, which is denoted by siy to be consistent in notation with the previous section. To
produce siy , activities s
jx in industry j = 1,…, N at each country s = 1,…,G are needed9. We first
need to know the levels of all gross outputs sjx associated with the production of s
iy . This is
estimated using the Leontief inverse as in equations (3) and (5).
9 Production stages in the global supply chain are identified by each s
jx , the maximum number of production stages of a specific supply chain in this accounting framework is G by N, assuming industries with the same classification but located in different countries produce differentiated products and so are located in different production stages of the global supply chain. Such an assumption is similar to the Armington assumption that has been widely used in CGE models for decades.
16
To be more specific to our current analysis, let us extend equations (3) and (5) to cover
any number of countries (G) and sectors (N). Then we obtain the following equations.
=
GGGGG
G
G
G Y
YY
BBB
BBBBBB
X
XX
2
1
21
22221
11211
2
1
(12)
=
∧
∧
∧
∧
∧
∧
∧
GGGGG
G
G
Gc
c
c
c
Y
Y
Y
BBB
BBBBBB
F
F
F
YBF
00
00
00
00
00
00
ˆ 2
1
21
22221
11211
2
1
=
∧∧∧∧∧∧
∧∧∧∧∧∧
∧∧∧∧∧∧
GGGGc
GGc
GGc
GGccc
GGccc
YBFYBFYBF
YBFYBFYBF
YBFYBFYBF
2211
2222221212
1121211111
(13)
With G countries and N sectors, A, B, ∧
F and Y are all GN×GN matrices. Bsr denotes the
N×N block Leontief (global) inverse matrix, scF is a 1 by N vector of direct emission intensities
in country s, placed along the diagonal of the GN by GN matrix of ∧
F . The subscript c represents
type of energies and non-energies. Five types are considered: (1) coal, (2) petroleum, (3) gas, (4)
waste, and (5) others (non-energy). ∑=G
r
srs YY is an N×1 vector that gives the global use of final
goods produced by s. Each column of the YB ˆ matrix of Equation (13) is a GN by 1 vector, the
number of non-zero elements in such a column vector represents the number of production stages
in our accounting framework for the global supply chain of a particular final good or service sjy .
Based on equation (13), we can decompose the total emissions of a final good or service
by production stages and types of energy in a global supply chain based on backward industrial
linkage as follows.
17
∑≠
∧∧
+=G
sr
srsrc
ssssc
sc YBFYBFYP )( for c =1,2,3,4,5 (14)
∑=
=5
1)()(
c
sc
s YPYP (15)
The first term in equation (14) consists of the diagonal elements in the last matrix of
equation (13), representing emissions generated in domestic production process; while the
second term in equation (14) is the sum of off-diagonal elements across the row and in a column
in the last matrix of equation (13), measuring emissions generated in foreign production
processes. The summation in the second term indicates that these emissions generated by foreign
production can be further split according to their source countries. Note that s
c
sc FF∑
=
=5
1
, that is,
emission intensities by energy types in each country/industry sum to the total emission intensity
of that country/industry. Therefore, equation (15) measures the total global emissions for the
production of final products in country s. The decomposition of total emissions by the production
of a final products in a global supply chain based on backward industrial linkage made by
equations (14) is shown in Figure 2.
Based on equation (14), the consumption-based national emissions inventories for a
particular product riy can be estimated for each country as a sum weighted by consumption
source structure:
∑=G
s
sicr
i
srir
iconsumer
c yPyy
yP )()( **
* for c =1,2,3,4 5 ; i=1,2,…N (16)
Here, ∑=G
r
sri
si yy * is the total final production in country s of product i for all countries, and
∑=G
s
sri
ri yy* is the total final consumption in country r of product i sourced from all countries.
Using the estimates from equation (14) and weighting by each country’s source structure
of the particular products it consumes, equation (16) allows one to estimate consumption-based
emissions at country/product level and its results are different from emissions estimates obtained
18
by using production emissions minus exported emissions plus imported emissions. Taking
automobile as an example, the production plus net transfer method widely used in the literature
only can provide estimates on how much of the emissions produced in the global auto industry is
consumed in a country, which does not equal global emissions induced by the total automobile
consumption in that country. However, summing over all products or industries, the total
consumption-based emissions for a country will be the same regardless backward or forward
linkage based computation is used.
Figure 2 GHG emissions in global supply chains – backward industrial-linkage-based decomposition
GHG emissions generated by foreign
segment of GVC
Total GHG emissions to produce final goods and
services
GHG emissions generated by domestic
segment of GVC
From Coal
From Petroleum
From Gas
From Waste
From Coal
From Petroleum
From Gas
FromWaste
By emission sources
From different source countries
From Non-
Energy source
From Non-
Energy
source
19
2.4 Measures of embodied emissions in trade by various GVC routes and their role in linking production-based and consumption-based emissions accounts
In recent years, the international trade of embodied emissions has been a subject of substantial
interest in both academic and policy circles. However, most MRIO-based measures of trade in
embodied emissions in the literature have not made a clear distinction between emissions
calculated by forward versus backward industrial linkages and often focus on the global and
country aggregate level. As we will show in this section, such a distinction is not important at an
aggregated level, but is crucial at a disaggregated level.
2.4.1 Forward industrial linkage based emission trade measures
At a bilateral sector or country sector level, emissions exports based on forward industrial
linkages (we labeled as EEX_F) for sector i and region s, are the emissions generated in sector i
to produce, directly and indirectly, gross exports from s to any other destination country except
country s itself (e.g., emission exports from the US chemical sector would include emissions
embodied in US steel and machinery sectors in addition to emission embodied in the US
chemical sector). There are two key issues to highlight here. First, using the example of
emissions exports from the US chemical industry, is that some of the emissions produced by that
sector can be exported indirectly via other US sectors such as steel, because US produced
chemicals are used as intermediate inputs in the production of steel exports. Second, the portion
of the emissions that is associated with products first exported but eventually re-imported to
satisfy domestic final demand is not part of the embodied emissions exports.
Emissions embodied in a country’s gross exports, which we labeled as EEG, refer to
emissions generated from the production of the country’s gross exports. Because this measure
focuses only on where the emissions come from but not where they are absorbed, it does not
exclude the part of the emissions that is generated by producing intermediate inputs for other
countries but eventually returns home via imports (i.e., is re-imported) to satisfy domestic final
demand. It is conceptually similar to emissions embodied in bilateral trade (EEBT) defined by
Peters (2008) and Peters et al. (2011). The EEG based on forward industry linkage, EEG_F,
20
refers to the part of emissions generated from the production of the country’s gross exports from
all sectors that originated from a particular domestic sector, including the portion that eventually
returns (which will be labeled REE_F) via imports. Because we already have a complete
decomposition of emissions by industry in equation (11), it is convenient to mathematically
specify EEX_F, emissions generated in production to satisfy foreign final demand, and REE_F,
emissions generated in the production of intermediate exports for other countries which are then
used to produce their exports and shipped back to country s as follows.
trG
rst
strrsrsrsssr YBFYBFYBFFEEX sss ∑≠
∧∧∧
++=,
_ (17)
ssrssrsstsG
rst
rtsrssrsrrsrsstsG
t
rtsrsssr YBALFYBALFYBALFYBALFFREE ssss∧∧∧∧
++== ∑∑≠ ,
_ (18)
Equation (17) is the sum of the third and fourth terms in equation (11) plus an additional
term taken from the last term of equation (11) which only sums over third country t re-exports to
a particular trading partner r (without the second summation over all r). Equation (18) is a
further decomposition of the second term in equation (11). It measures domestic emissions
embodied in intermediate exports from country s to country r that return to s and are ultimately
absorbed in s via all possible routes through forward industrial linkage. Both portions are
emissions related to international trade but for different market segments.
We specify domestic emissions embodied in gross exports from country s to country r
based on forward industrial linkages as
)2(
,
)2(
)2()1(
_
c
G
rst
strssrG
sr
ttrtsrss
b
G
t
tsrtsrss
a
G
t
trrtsrsssrsssrsssr
YBAYBALFYBALF
YBALFYLFELFFEEG
ss
sss
+++
+==
∑∑∑
∑
≠≠
∧∧
∧∧∧
(19)
It measures what amount of domestic emissions can be generated from the production of gross
exports srE in country s, regardless whether these gross exports are finally absorbed in
importing country r or not. It can be decomposed into two parts:
21
1. Domestic emissions generated from the production of final goods exports,
2. Domestic emissions generated from the production of intermediate goods exports that
are:
2a. finally absorbed in the direct importing country r,
2b. returned (re-imported) to the exporting country s, or
2c. re-exported to a third country t.
It is identical to the “Emissions Embodied in Bilateral Trade” (EEBT) defined by others (Peters
2008; Peters and Hertwich 2008) in the literature on embodied emissions in trade. It is easy to
see that REE_Fsr defined by equation (18) is exactly the third term in equation (19). We can
show that, at the bilateral-sector level, )__( srsrsrss FREEFEEXELF s +≠∧
due to indirect
emissions exports through third countries. However, after aggregating over all trading partners,
at the country-sector level,
∑∑∑≠≠≠
∧
=+=G
sr
srssG
sr
srsrG
sr
sr ELFFEEXFREEFEEG s)__(_ (20)
The step by step derivation of equations (18) to (20) can be found in appendix A.2. The intuition
behind the derivation is simple: both srFEEX _ and srFREE _ require that the emissions
associated with a product is consumed in destination country r by definition, while srFEEG _ or
EEBT do not have such restrictions and are concerned only where these emissions are generated,
regardless of where their associated products are finally absorbed.
Similar to Peters et al. (2011), we define the balance of embodied emissions in trade, or
“net emissions transfer” as
∑∑≠≠
−=G
rs
rsG
sr
srs FEEXFEEXT __ (21)
It is easy to show that sT equals the difference between production-based and consumption-based
emission inventory. That is,
)()( ri
consumerri
producererr yPyPT −= . (22)
2.4.2 Backward industrial linkage based emission trade measures
22
Embodied emissions exports calculated by backward industrial linkages at a bilateral
sector or country-sector level, which we labeled as EEX_B, refer to the amount of emissions
generated by the production of a particular sector’s gross exports (e.g., US auto), which will
include emissions produced by any domestic sectors (e.g., including US rubber, chemicals, steel,
and glass) via backward industrial linkages, and is ultimately absorbed abroad or in a particular
destination country. There are also two key features to take into account. First, the measure
quantifies emissions to the sector whose products are exported. Second, the concept excludes the
part of domestic emissions that is eventually re-imported. In general, at the country sector and
bilateral sector level, EEX_F and EEX_B are not the same except by coincidence. However,
once we aggregate across all sectors, the distinction between EEX_F and EEX_B disappears.
To trace emissions generated by gross trade flows at bilateral and sector levels, it is useful
to think of the total domestic emissions associated with gross trade flows that is absorbed abroad,
denoted by EEX, as a distinct concept from EEX_B or EEX_F in order to measure emissions
embodied in a particular bilateral gross trade flows. It is also based on backward industrial
linkages and is also ultimately absorbed abroad, similar to EEX_B, but does not require
domestically produced emissions to be absorbed in a particular destination country. In other
words, at the country sector level, this third trade-in-emissions measure is the same as EEX_B,
but at the bilateral or bilateral sector level, they are different. As we will show later in this paper,
EEX is the only emissions trade measure that is consistently associated with bilateral gross trade
flows, while both EEX_F and EEX_B are not, due to indirect emissions trading through third
countries. All these three measures exclude the part of domestic emission that first exported but
eventually returns home. However, all of them are useful to trace emission trade in gross exports
for different purpose beyond the country aggregate level. For instance, if one wishes to
understand the global emissions level generated by a country’s gross exports and its source
structure, the backward-linkage-based emissions measures are the right one to use. If one wishes
to understand the responsibility for emissions from a given sector in the country’s gross exports
from all sectors, one should use the forward-linkage-based measures.
As we have already shown, to decompose a country/industry’s total GHG emissions by
source of final demand and measure domestically produced emissions embodied in a country’s
gross exports from all sectors based on forward industrial linkage, applying Leontief’s original
23
method is sufficient. However, for measuring global emissions generated by a country’s gross
exports and tracing its source structure based on backward industrial linkage, Leontief’s original
method will not be sufficient, as it does not provide a way to decompose gross intermediate trade
flows across countries according to their final absorption, as illustrated by Wang et al. (2013) in
their recent work.
Following Wang et al.’s innovative intermediate trade flow decomposition method, we
define our bilateral emissions trade measures based on backward industrial linkage as
+++
+=
∑ ∑∑∑≠ ≠≠≠
G
rst
G
tsu
turtsrttG
rst
rtsrG
rst
rtrrsrTsss
rrrrsrTssssrTssssr
YBAYBAYBALF
YBALFYBFEEX
, ,,,)()()(#)(
)(#)(#)( (23)
+++
+=
∑∑∑∑≠ ≠≠≠
G
rst
G
tsu
urtusttrG
rst
rtsrG
rst
trttstTsss
rrrrsrTssssrTssssr
YBAYBAYBALF
YBALFYBFBEEX
, ,,,)()()(#)(
)(#)(#)(_ (24)
where “#” is an element-wise matrix multiplication operator10. To facilitate the understanding of
the three terms in the emissions trade measure defined in equation (23), we provide the following
intuitive interpretations.
The the 1st term, srTsss YBF #)( , represents domestic emissions generated by the
production of final exports from country s to country r. The 2nd term, )(#)( rrrrsrTsss YBALF ,
represents domestic emissions generated by the production of intermediate exports from country
s used by direct importer (country r) to produce final goods and services which are consumed in
country r. The 3rd term, #)( TsssLF {…} represents domestic emissions generated by the
production of intermediate exports from country s used by the direct importer (country r) to
produce intermediate or final goods and services that are re-exported to a third country t. The
three elements in the parenthesis, ∑≠
G
rst
rtrrsr YBA,
, ttG
rst
rtsr YBA ∑≠ ,
, and ∑∑≠ ≠
G
rst
G
tsu
turtsr YBA, ,
show how the
re-exports are produced in country r by using intermediate exports from country s as inputs.
10For example, when a matrix is multiplied by 1×n column vector, each row of the matrix is multiplied by the corresponding row element of the vector.
24
They represent final goods re-exports, intermediate goods re-exports for third countries’
domestically consumed final goods, and intermediate goods re-exports for third countries’ final
goods exports, respectively.
It is interesting to note that the difference between srEEX (23) and srBEEX _ (24)
appears in only the third country term (the third term). The former includes emissions absorbed
not only by country r, but also by third countries t and u (last three terms in equation 24). The
latter includes not only emissions exports from country s embodied in its own gross exports to
country r (the 1st and 2nd terms in equation 24, which are the same as the first two terms in
equation 23), but also emissions exports by country s embodied in its gross exports to third
country t, that are finally absorbed by country r (the last terms in equation 24). This illustrates
why we claim that srEEX is the only measure of emission trade which is consistently associated
with bilateral gross trade flows. Both emissions export measures are deviate from gross bilateral
trade flows due to indirect trade through third countries.
Similar to the definition of EEG_F, we could also define EEG_B, the measure of
domestic emissions generated from the production of bilateral gross exports at sector level based
on backward industrial linkage, which refers to emissions from all domestic sectors induced by
the production of a particular sector’s gross exports to a particular trading partner or the rest of
the world, including the portion of emissions associated with exported products that are
eventually re-imported, REE_B.
+++
+==
∑∑∑
∑
≠≠
G
rst
strssrG
sr
ttrtsrTsssG
t
tsrtsrTsss
G
t
trrtsrTssssrTssssrTssssr
YBAYBALFYBALF
YBALFYLFELFBEEG
,#)(#)(
#)(#)(#)(_ (25)
EEG_Bsr measures what amount of domestic emissions can be generated from all sectors in
country s in the production of gross exports srE , regardless of whether these exports are finally
absorbed in importing country r or not. The four terms in equation (25) have similar
interpretations to those of the four terms in equation (20); the differences are that these terms
include not only domestic emissions generated by the exporting sectors, but also those of other
upstream domestic sectors that contribute to the production of a particular sector’s gross exports.
25
We define emissions embodied in intermediate exports that are first exported but
ultimately returned and absorbed at home based on backward industrial linkages from country s
to country r as:
)(#)()(#)()(#)(
#)(_
,
ssrssrTssstsG
rst
rtsrTsssrsrrsrTsss
tsG
t
rtsrTssssr
YBALFYBALFYBALF
YBALFBREE
++=
=
∑
∑
≠
(26)
It can be seen that REE_Bsr is exactly the third term in equation (25). We can show that EEG_Bsr
equals the sum of equations (23) and (26) at the country aggregate level only.
∑∑∑≠≠≠
=+=G
sr
srsssG
sr
srsrG
sr
sr ELFBREEEEXuBuEEG )_(_ (27)
where, u is a 1 by N unit vector. Detailed proofs of equations (25) to (27) are given in appendix
A.3.
To completely measure total emissions from the production of a country’s gross exports,
emissions generated in other countries that provide intermediate inputs for the exporting country
also have to be estimated. The foreign-produced emissions embodied in a country’s gross exports
(FEE) can be defined as
)(#)(#)(
)(#)(#)(
,,
rrrrsrTG
rst
tstsrTG
rst
tst
rrrrsrTrsrsrTrsrsr
YLABFYBF
YLABFYBFFEE
∑∑≠≠
++
+= (28)
Each term in equation (28) has an intuitive interpretation. The first term, srTrsr YBF #)( , is
the importer’s (country r) emissions embodied in the final exports of country s to country r. The
second term, )(#)( rrrrsrTrsr YLABF , is the importer’s emissions embodied in the intermediate
exports of country s to country r, which are then used by country r to produce its domestic final
goods and services. The third term, srTG
rst
tst YBF #)(,∑≠
, is foreign emissions from third countries t
embodied in the final exports of country s to country r. The last term, )(#)(,
rrrrsrTG
rst
tst YLABF∑≠
, is
26
foreign emissions from third country t embodied in the intermediate exports of country s to
country r, which are then used by country r as inputs to produce its domestic final goods and
services.
Combining equations (23), (26) and (28), we decompose the total global emissions
generated from the production of a country’s gross exports to its trading partner as
)8()7()6()5(
)(#)(#)()(#)(#)(
)4()3(
#)()()()(#)(
)2()1()(#)(#)()(
,,
, ,,,
rrrrsrTG
rst
tstsrTG
rst
tstrrrrsrTrsrsrTrsr
tsG
t
rtsrTsssG
rst
G
tsu
turtsrttG
rst
rtsrG
rst
rtrrsrTsss
rrrrsrTssssrTssssr
YLABFYBFYLABFYBF
YBALFYBAYBAYBALF
YBALFYBFEP
∑∑
∑∑∑∑∑
≠≠
≠ ≠≠≠
++++
+
+++
+=
(29)
The first four terms of equation (29) represent emissions within the exporting country,
which are a by-product of generating the exporting country’s GDP; the last four terms in
equation (29) represent emissions within foreign countries that provide intermediate inputs for
the exporting country, but also create GDP for these foreign countries. The decomposition made
in equation (29) is also shown in Figure 3.The number in the lowest level box corresponds to the
terms in equation (29).
Figure 3 Decomposition of global GHG emissions in the production of gross exports by different GVC routes – based on backward industrial-linkage
Global GHG emissions in production of a country’s gross
exports
Domestic GDP and GHG emissions in producing exports
Foreign GDP and GHG emissions in supplying imported inputs
Consumed by direct
importers- final goods
Consumed by direct
importers- intermediate
goods
Consumed in third
countries
Consumed in exporting
country
Embedded in imports
In final goods,
produced and
consumed by direct importer
In intermediate
goods, produced and consumed by
direct importer
In final goods, produced by
third countries consumed by
direct importer
In intermediate goods,
produced by third countries consumed by
direct importer
Who produces it?
Who consumes it?
27
2.4.3 Relationships among different emissions trade measures
It turns out that separating emissions by backward versus forward industrial linkages is
crucial to properly tracing emissions in trade at a disaggregated level. To our knowledge, the
literature on embodied emissions in trade has not previously made a clear distinction between
them. While Peters et al. (2011) made a distinction between emissions embedded in bilateral
trade (EEBT) versus embodied emissions of final consumption, they do so only at the country
aggregate level. More importantly, they do not distinguish backward from forward industrial
linkages—such a distinction is not important at the country aggregate level, but is crucial at a
disaggregated level. Therefore, a key contribution of this paper is to systematically develop these
quantitative emissions trade measures at both aggregated and disaggregated levels. The
relationships among these different emissions trade measures can be summarized as follows:
In a world of three or more countries, domestic emissions generated by the production of
bilateral gross exports to satisfy foreign final demand (EEX), forward linkage-based emissions
exports (EEX_F), and backward linkage-based emissions exports (EEX_B) are, in general, not
equal to each other at the bilateral/sector level, though they are the same at the country aggregate
28
level. EEX_F and EEX_B are also equal at the bilateral aggregate level, while EEX and EEX_B
are the same at the country/sector level.
EEG_F and (EEX_F + REE_F) are equal to each other at both country sector and country
aggregate levels, but not equal at the bilateral sector level; while EEG_B and (EEX_B+ REE_B)
are equal to each other only at the country aggregate level. Because both REE_F and REE_B are
non-negative, EEG_F is always greater than or equal to EEX_F at country/sector level; both
EEG_F and EEG_B are always greater than or equal to all the three measures of trade in
embodied emissions (EEX, EEX_F and EEX_B) at the country aggregate level. While at the
bilateral sector level, EEG (EEBT) measures can greater or smaller than EEX measures, as
discussed in detail by Peters (2008). Finally, EEX_F and EEG_F as well as (EEX_F+REE_F)
are always less than or equal to the sector-level total emission production )( siyP .
The intuition behind these statements is simple: since direct emissions exports at the
sector level are the same for all three trade-in-emissions measures, only indirect emissions trades
may differ. However, because such indirect emissions exports are part of the total emissions
produced by each sector, the total emissions in a country/sector set an upper bound for forward
linkage-based emissions exports and domestic emissions embedded in gross exports.
The definition of all the embodied emission trade measures discussed in this section and
their relationships are summarized in Tables 1a and 1b below:
29
Table 1a Definition of different measures of embodied emissions in trade
Acronym
or label
Definition in words Key characters
Equation #
in text
EEX_F Embodied emissions exports,
forward-linkage-based
1. Emissions generated in producing goods and
services that satisfy foreign final demand;
2. Include indirect emissions exports ;
3. Excluding emissions associate with intermediate
exports that are returned and absorbed at home
4. Trade concepts, produced in one country, consumed
by another.
17
EEX_B Embodied emissions exports,
backward linkage –based
24
EEX Embodied emissions
associated to gross bilateral
trade flows
23
REE_F Embodied emissions return
home, forward linkage–based
Emissions generated by producing intermediate inputs
exported to other countries, which eventually returns
home via imports to satisfy domestic final demand
18
REE_B Embodied emissions return
home, backward linkage–
based
26
EEG_F Emissions embodied in a
country’s gross exports,
forward linkage-based
1. Production concept, consistent to GDP by industry
statistics
2. Focuses only on where the emissions are produced
3. Include the part of emissions that is generated by
producing intermediate inputs for other countries but
eventually re-imported
19
EEG_B Emissions embodied in a
country’s gross exports,
backward-linkage-based
25
Table 1b Relationships among different measures of embodied emissions in trade
Aggregation level
EEX &
EEX_F
EEX &
EEX_B
EEX_F &
EEX_B
REE_F &
REE_B
EEG_F &
EEG_B
EEG_F &
(EEX_F+ REE_F)
EEG_B &
(EEX_B+ REE_B)
srie Bilateral-
Sector ≠ ≠ ≠ ≠ ≠ ≠ ≠
∑=
N
i
srie
1
Bilateral Aggregate
≠ ≠ = = = ≠ ≠
∑≠
G
sr
srie
Country-Sector
≠ = ≠ ≠ ≠ = ≠
∑∑≠ =
G
sr
N
i
srie
1
Country Aggregate
= = = = = = =
30
31
3. Empirical analysis
Following the concepts and accounting framework proposed above, this section uses the
WIOD11 to demonstrate how this framework can help to gain a deeper understanding of the
relationships between GVCs and CO2 emissions from different perspectives. While we focus on
CO2 here, the framework works in the same way for any environmental stressor.
3.1 Tracing CO2 emissions in GVCs at the national level We first apply the accounting framework at the national level to demonstrate the concepts
summarized in Figures 1, 2, and 3.
Figure 4 shows “who produced CO2 emissions for whom” by different GVC routes in
2009, using the two largest emitters, China and the US, as an example. This figure follows the
forward industrial-linkage-based downstream decomposition method (Figure 1). Clearly, most
CO2 emissions (EH_F) are the result of satisfying the domestic final demand in each country that
not relate to international trade. This result holds for most large economies since the self-
sufficient portion normally accounts for the largest part of total final demand. However,
compared to the US, this portion is much lower in China. More than 30% of China’s CO2
emissions are induced by foreign final demand (EEX_F=EEX_F1+EEX_F2+EEX_F3). This is
mainly for two reasons: 1) after China’s accession to the WTO, foreign final demand has played
an increasing role in driving the growth of China’s GDP and the generation of China’s CO2
emissions (Peters et al. 2011); 2) the CO2 emission intensity for producing one unit GDP in
China is higher than that in the US (Davis and Caldiera 2010) (also see Appendix B4).
As we discussed in section 2, part of the CO2 emissions induced by domestic final
demand depend on international trade due to production sharing between home and foreign
countries, measured by REE_F. As an example, producing a car in China to satisfy China’s own
final demand may require the importation of an engine from the US, which may use Chinese
metal parts as inputs in its production. As a result, China’s final demand for its domestic final
products may cause its own CO2 emissions to rise through the two-way international trade in
11 www.wiod.org
32
intermediate goods and services. The forward industrial-linkage-based downstream
decomposition method can also be used to trace foreign final demand in driving home-country
produced CO2 emissions by different GVC routes. As also shown in Figure 4, the share of CO2
emissions induced by foreign final demand through final goods trade (EEX_F1) for China is
obviously larger than that for the US. This depends on both the CO2 emission intensity and how
a country participates in GVCs. Most developing countries, such as China, join GVCs through
exporting relatively large amounts of final products in their early stage of development.
Figure 4 Who produces emissions for whom (forward industrial-linkage-based decomposition, 2009)
Figure 5 uses Germany and China as an example to show how CO2 emissions are
generated from upstream production stages in GVCs by different emission sources when these
two countries produce final goods and services. This figure follows the backward industrial-
linkage-based upstream decomposition method (Figure 2). The foreign emissions induced by the
production of final goods and services in Germany account for a relatively large share (more
than 35% in 2009) compared to that in China (less than 10% in 2009). This depends not only on
all related countries’ CO2 emission intensities, but also their cross country production sharing
arrangements and the way they participate in GVCs. China’s CO2 emission intensity is higher
0% 25% 50% 75% 100%
CHN
USA
For domestic final demandFor foreignfinal demand
USA
China
EH_F: CO2 emissions for domestic final demand without through international trade
REE_F: CO2 emissions for domestic final demand through international trade (feedback)
EEX_F1: CO2 emissions for trade partner's final demand through final goods trade
EEX_F2: CO2 emissions for trade partner's final demand through intermediate goods trade
EEX_F3: CO2 emissions for trade partner's final demand through intermediate goods trade by way of third countries
33
than that of Germany (see Appendix B4); this makes China’s domestic emissions take a
relatively large share in the production of final goods. On the other hand, Germany’s value chain
has a relatively large foreign segment (relative to China, a country which is less integrated into
the European Union), so more emissions may occur in other countries due to the induced
demand for intermediate imports used for producing German-made final products.
In addition to technological efficiency, the amount of induced CO2 emissions when
producing final products may also depend on the structure of energy use in upstream production
processes. For example, the usage of coal accounts for a very large portion of domestic
emissions for China and relatively large portion of foreign emissions for Germany when
producing final goods and services. In general, this indicator can help us clearly understand how
a country’s production of final goods and services impact on the CO2 emissions in its upstream
countries or industries (domestic or foreign) through various GVC routes.
Figure 5 Induced emissions in both domestic and international segments of GVC when a country produces final goods and services (backward industrial-linkage-based decomposition, 2009)
Figure 6 shows how Japan and China’s gross exports generate both domestic and foreign
CO2 emissions by different GVC routes in 2009 (cf. Davis and Caldiera 2010). This figure
corresponds to the backward industrial-linkage-based decomposition of gross exports (Figure 3).
Compared to Japan, domestic CO2 emissions generated from China’s gross exports production
0% 25% 50% 75% 100%
CHN
DEU
Coal Petroleum Gas Waste Other
Domestic emissions Foreign emissions
China
Germany
34
account for a relatively large share (more than 90%). Though China imports more intermediate
inputs than Japan does in producing gross exports, lower energy efficiency and high carbon
intensity are the main drivers that increase China’s domestic emissions share in gross exports.
When looking at the domestic CO2 emissions by GVC routes, a remarkable difference between
Japan and China can be observed: Japan’s domestic CO2 emissions in gross exports are mainly
generated in the production of intermediate goods and services that are exported to its trading
partners, while, for China, final goods exports play a dominant role. This depends on both the
way a country participates in GVCs and its CO2 emission intensity. As a result of its
comparative advantage in assembly, exports final products is one of the major ways that China
participates in GVCs. While Japan participates in GVCs largely through high-tech intermediate
exports as a result of its comparative advantage in capital and skill intensive activities. Though
the major exports with high comparative advantage for China are textile and electrical products
which may not emit a large amount of CO2 in their production processes, domestic intermediate
inputs such as high-carbon electricity and chemicals are directly and indirectly embodied in
these final product exports. As a result, domestic CO2 emissions through final goods trade in
China accounts for a relatively large share of its total emissions induced by gross exports.
Figure 6 Emissions embodied in gross exports by eight GVC routes (backward industrial-linkage-based decomposition, 2009)
0% 25% 50% 75% 100%
CHN
JPN
route 1: domestic emissions consumed by trade partner countries through final goods traderoute 2: domestic emissions consumed by trade partner countries through intermediate goods traderoute 3: domestic emissions consumed by third countries through intermediate goods traderoute 4: domestic emissions consumed by itself through intermediate goods traderoute 5: trade partner country's emissions consumed by itself through final goods traderoute 6: trade partner country's emissions consumed by itself through intermedaite goods traderoute 7: third countries' emissions consumed by trade partner country through final goods traderoute 8: third countries' emissions consumed by trade partner country through intermediate goods trade
Domestic emissions in gross exportsForeign emissionsin gross exports
35
The share of foreign CO2 emissions in a country’s gross exports also depends on its
trading partners’ CO2 emission intensities. Japan’s import content in exports is lower than that of
China, but its foreign emissions in gross exports are higher. This implies that relatively high
foreign carbon intensity goods are embodied in Japan’s gross exports. In addition, one important
advantage of using this framework is that we can easily understand who produces gross exports
and CO2 emissions for whose consumption through which specific GVC route. For example,
about 20% of CO2 emissions in Japan’s gross exports is for satisfying its direct trading partner’s
final demand, but this is emitted in third countries through Japan’s use of third countries’
intermediate goods and services to produce its exports to the partner country (route 7 and 8).
Given the rapid extension of international fragmentation of production, this type of emissions in
international trade tends to increase if no global treaty is in place. We report more detailed
results on CO2 emissions based on the 3 type decomposition method discussed in section 2 at the
national level for the years between 1995 and 2009 in Appendix B1- B3.
3.2 Tracing CO2 emissions in GVCs at the bilateral and sectoral levels
As discussed in section 2, the unified accounting framework proposed in this paper can also be
used to trace CO2 emissions in GVCs at detailed bilateral and sectoral levels. Figure 7 shows
how emissions are generated in the CO2 intensive metal industry in three selected countries,
China, Mexico, and Poland, to satisfy US final demand through different GVC routes. This
figure corresponds to Figure 1 following the forward industrial-linkage-based decomposition
method. We use these three countries as an example here because they are all active players in
GVCs of metal products and are also important direct or indirect trading partners of the US,
while being located in three different continents: North America, Asia, and Europe. In addition,
for most countries, the metal industry is always one of the largest emitters, with relatively high
carbon intensity.
Figure 7 Metal industry's CO2 emissions exports from selected countries to the US by different GVC routes (forward industrial-linkage-based decomposition, 2009)
36
Figure 7 shows the CO2 emissions in the metal industries in these three countries from
activities to satisfy US’s final demand via different GVC routes. The pattern is mainly
determined by a country’s position and participation in GVCs. China exports large quantities of
final products to the US, so we see China’s metal industry’s CO2 emissions from satisfying US’s
final demand arising mainly through final goods trade. Mexico is also close to the US consumer
but unlike China, it is located in a relative upstream position in metal GVCs: it is one of the
largest providers of parts and components of metal products to the US, for example, for the US
auto industry. As a result, the CO2 emissions in Mexico’s metal industry are mainly embodied in
its export of intermediate goods which are directly and indirectly consumed in the US. Poland is
much further from the US consumers and is embedded in the EU economy, so it is located far
upstream in the GVCs of metal products. Therefore, a large portion of Poland’s metal industry
CO2 emissions are embodied in goods traded with third countries, such as metal products used in
a German car finally consumed in the US. Tracing CO2 emissions at the bilateral and sector
levels as this example can help us to better understand the effect of a country’s position and
participation in GVC on the geographic source of its CO2 emissions at the industry level.
Following the accounting method summarized in Figure 2, we use German-made and
Chinese-made cars as an example to demonstrate how these two large car producers cause
0% 25% 50% 75% 100%
Poland→USA
Mexico→USA
China→USA
EEX_F1: CO2 emissions for trade partner's final demand through final goods tradeEEX_F2: CO2 emissions for trade partner's final demand through intermediate goods tradeEEX_F3: CO2 emissions for trade partner's final demand through intermediate goods trade by way of third countries
37
upstream country’s CO2 emissions in automobile GVCs. Figure 8 shows China, the rest of the
world (RoW), and Russia are the economies most affected by car production in Germany,
besides Germany itself. On the one hand, this is because these three economies are located
upstream of Germany’s car value chain through providing intermediate goods and services
directly or indirectly for German car production. On the other hand, it is a result of the relatively
high carbon intensity for producing intermediate goods in these countries compared to other
upstream countries, like the US and Japan. Another important factor is that different upstream
countries involved in Germany’s car value chain rely on different energy sources to produce
their intermediate exports. For instance, China mainly relies on coal-based energy, hence coal-
based CO2 emissions account for the majority of emissions in China resulting from car
production in Germany. This also implies that emissions to produce German cars will decrease
substantially if China can replace coal by other green energy sources in producing intermediate
goods purchased by the Germans. Compared to the German-made car, the production activities
of auto makers in China have a larger impact on CO2 emissions in the RoW and Russia. China
overtook the US, becoming the world’s top auto maker and market in 200912. Large amounts of
components are imported from the RoW through various GVC routes directly and indirectly. As
a result, the RoW has been the most affected upstream region in the production of Chinese-made
cars. In addition, Japan and the US are also heavily affected since both countries are located in
the upstream of China’s car value chain by providing high-tech intermediate goods and services.
This is different from the cars made in Germany because Germany may obtains almost all high-
tech parts from its domestic suppliers rather than its main rivals, the US and Japan.
Figure 8 Induced foreign CO2 emissions from producing cars in selected countries (backward industrial-linkage-based decomposition)
12 China Daily, http://www.chinadaily.com.cn/bizchina/2010-01/12/content_9309129.htm, Updated: 2010-01-12 15:37
38
To illustrate how the accounting framework proposed in Figure 3 works at bilateral and
sector levels, we use Germany, Mexico and China’s electrical product exports to the US as an
example. Figure 9 demonstrates how a country’s gross exports of electrical products to the US
generate both domestic and foreign CO2 emissions through different GVC routes. These three
countries were the largest trading partners for electrical products with the US in Europe, North
America and Asia, respectively, in 2009. Figure 9 shows that about 85% of CO2 emissions
generated by China’s gross exports of electrical goods to the US are emitted inside China, a very
large portion of which is from the production of final goods exported to the US. Compared to
China, Germany and Mexico show a very different pattern. Their exports of electrical product to
the US induce more foreign CO2 emissions. This difference is caused by several reasons that
may operate in opposing directions: for instance, a higher domestic carbon intensity in producing
goods and services leads to a larger portion of domestic emissions; while a higher proportion of
foreign intermediate imports in a country’s exports (implying a higher participation in GVCs),
leads to a smaller portion of domestic emissions.
Figure 9 CO2 emissions embodied in selected countries’ gross exports of electrical products shipped to the US via 8 different GVC routes (backward industrial-linkage-based decomposition, 2009)
0 2500 5000 7500 10000
China
India
Japan
USA
GBR
DEU
RUS
RoW
Coal Petroluem Gas Waste Other
025005000750010000
China
India
Japan
United States
United Kingdom
Germany
Russia
the rest of the world
When cars are produced in Germany When cars are produced in China
39
Estimates based on WIOD shows that the import contents of electrical product exports to
the US are 24%, 53% and 32% for Germany, Mexico and China, respectively. Germany’s import
contents are the lowest of these three exporting countries, but its gross exports to the US
generate more foreign CO2 emissions. This clearly reflects two factors. First, Germany has
relatively low domestic carbon intensity in producing exports. Second, Germany may import
more high-carbon intensity intermediate goods directly or indirectly from other countries for
producing its gross exports to the US. Mexico’s imported content in its exports is the highest.
This naturally leads to a large portion of foreign CO2 emissions in its gross exports. The US’s
CO2 emissions generated by gross exports of electrical products from Mexico to the US accounts
for a very large portion (routes 5 and 6) compared to that in other countries. This is mainly
because Mexico needs more intermediate parts and components provided by the US directly or
indirectly when producing electrical products for exporting back to the US. In addition, this
accounting framework not only identify who produces gross exports and CO2 emissions, but also
identify who finally consumes the CO2 emissions embodied in the gross exports. Clearly, the
embodied CO2 emissions in routes 1, 2, 5, 6, 7, and 8 are finally consumed by the US; emissions
in route 3 are finally consumed by third countries, emissions in route 4 are finally consumed by
0% 25% 50% 75% 100%
China→USA
Mexico→USA
Germany→USA
Domestic emissions in gross exports Foreign emissions in gross exports
route1: domestic emissions consumed by trade partner countries through final goods traderoute 2: domestic emissions consumed by trade partner countries through intermediate goods traderoute 3: domestic emissions consumed by third countries through intermediate goods traderoute 4: domestic emissions consumed by itself through intermediate goods traderoute 5: trade partner country's emissions consumed by itself through final goods traderoute 6: trade partner country's emissions consumed by itself through intermedaite goods traderoute 7: third countries' emissions consumed by trade partner country through final goods traderoute 8: third countries' emissions consumed by trade partner country through intermediate goods trade
40
the exporting countries themselves. The above example shows that border carbon adjustments
would be difficult because emissions could be embodied in gross exports through different
routes in GVCs due to different production sharing arrangements.
3.3 Bilateral Trade in CO2 Emissions
Figure 10 shows the bilateral trade in CO2 emissions across the 15 largest countries or country
groups for 1995 and 2009. In 1995, China, the US, EUW (the EU15), Russia and the RoW are
the major exporters of CO2 emissions; Japan, the US, the EUW and the RoW are the major
importers of CO2 emissions. The basic direction of bilateral flows remains unchanged between
1995 and 2009, but some interesting changes in the magnitude of CO2 emissions trade can be
observed. For example, China’s exports of CO2 emissions increased dramatically and, at the
same time, China also became one of the largest importers of CO2 emissions. More interesting
thing is that the carbon emission trade (exports + imports) between China and other developing
countries has exceeded all bilateral emission trade between any developed economy blocks and
China (the EU-China or the US-China). This is not only driven by the increased demand for
Chinese manufacturing products from developing countries, but also due to “made in China” is
highly depend on intermediate imports from other developing countries as inputs, and the RoW
uses more and more intermediate imports from China, both of them have much higher carbon
intensity than intermediate imports from developed countries. This could be a great concern
since both China and countries in the RoW are Non-Annex B economies in Kyoto Protocol and
have relatively weak environmental regulations.
41
Figure 10 Bilateral trade in CO2 emissions
1995
2009
42
Note: The magnitudes of emissions trade flows in this figure are based on EEX_Fsr. Exports from CHN (China) to the RoW (rest of the world) are respectively 104,563 Kt and 584,219 Kt for 1995 and 2009.
43
3.4 The relationship between GVC participation and embodied CO2 emissions in gross exports
As mentioned in previous sections, a country’s gross exports can generate both domestic and
foreign CO2 emissions through various GVC routes. The magnitudes of these two types of
emissions highly depend on a country’s position and participation in GVCs. The international
economics literature on vertical specialization indicates that a country could join GVCs in two
ways: it can participate in GVCs from downstream, use imported intermediate inputs to produce
exports, or from upstream, exports intermediate goods that are used as inputs by another country
to produce goods for exports. To determine a county’s position in a vertical integrated
production chain need both measures (Koopman et. al. 2014). Figure 11a shows the relationship
between a country’s GVC participation from downstream (similar to Hummels et al. (2001)’s
vertical specialization share indictor labeled as VS, measures the value of imported contents
embodied in a country’s exports) and its domestic share of total CO2 emissions embodied in
gross exports for the top 20 exporting economies in the world in 2009. The size of a bubble
represents the magnitude of foreign CO2 emissions embodied in a country’s gross exports. The
dark the color of the bubble, the higher the emission intensity (environment cost for per unit
GDP; emissions in KT / GDP in million US$ at 1995 constant prices). The rings with different
colors surrounding the bubbles show four different GVC routes (through energy, non-energy
final goods trade, energy, non-energy intermediate goods trade). The main facts revealed by
Figure 11a can be summarized as follows.
1. The higher the imported content in a country’s exports, the smaller the domestic CO2
emissions in its gross exports (ceteris paribus). When a country uses more foreign
intermediate inputs to substitute for domestic inputs in producing exports, relatively less
CO2 emissions will be generated domestically13. The large scale of gross exports
produced by China and the RoW and their relatively higher imported contents in exports
compared to similar large countries, such as the US and Japan, cause more foreign CO2
emissions. However, the relatively higher carbon intensity for developing economies,
like China, India and the RoW, leads to a larger share of domestic CO2 emissions
13 Without considering the energy goods trade, the level of GVC participation for the RoW should be much lower.
44
embodied in their gross exports, although their shares of imported contents in exports are
similar to some developed economies, such as Germany, France and Spain.
2. Developing economies join GVCs by providing relatively more final goods, which is
different from developed economies due to their different comparative advantages. For
example, the foreign CO2 emissions embodied in gross exports from the US, Japan,
Korea and Taiwan are mainly as a result of intermediate goods trade, while for China,
India and Mexico they are mainly as a result of final goods trade.
3. China and RoW have been the top two regions inducing massive foreign CO2 emissions
in producing exports. Besides their large scale of gross exports, both economies import
high-carbon intensity components from each other. While Japan, Korea and Taiwan’s
bubbles are not only relatively large but also darker (higher carbon intensity). This is
mainly because China has been their major trading partner, providing not just final goods
but also intermediate goods.
Figure 11a The relationship between GVC participation and CO2 emissions (2009)
45
Figure 11b shows the relationship between a country’s GVC participation from upstream
(similar to Hummels et al. (2001)’s vertical specialization share indictor labeled as VS1,
measures intermediate exports sent indirectly through other countries) and its domestic share of
total CO2 emissions embodied in gross exports. The horizontal axis remains no change, but
countries’ positions show very different pattern compared to that in Figure 11a. For example,
because developed economies, such as the US, Japan, UK, Germany and Taiwan can provide
more sophisticated manufacturing intermediates to their downstream countries for further
processing and assembling, thus have higher degree of GVC participation from upstream, while
India, Mexico and China have lower levels of participation. Viewing a country’s participation
from both upstream and downstream perspective provide more insights on the relationship
between GVC participation and emissions in trade. For instance, Korea and Taiwan’s positions
are very close in Figure 11a, but very different in Figure 11b.
RoWCHNDEU
USA
KOR
NLD
FRA
JPN
ITA
BEL
TWN
CAN
GBR
MEX
IND
ESP
SWEPOL
CZE
AUT
5%
15%
25%
35%
45%
45% 55% 65% 75% 85% 95%
GV
C p
artic
ipat
ion
degr
ee fr
om d
owns
trea
m
Domestic share of total CO2 emissions embodied in gross exports
2009
: 30 Million Ton Foreign CO2 emissionsembodied in gross exports
by the way of non-energy final goods trade by the way of non-energy intermedaite goods tradeby the way of energy-related final goods tradeby the way of energy-related intermediate goods trade
: Embodied foreign carbon intensity 1.00 Kt/Million US$: Embodied foreign carbon intensity 0.75 Kt/Million US$: Embodied foreign carbon intensity 0.60 Kt/Million US$
VS share
46
Figure 11b The relationship between GVC participation and CO2 emissions (2009)
RoW
CHN
DEU
USA
KORNLDFRA
JPN
ITA
BEL
TWN
CAN
GBR
MEXIND
ESPSWE
POL
CZE
AUT
5%
10%
15%
45% 55% 65% 75% 85% 95%
GV
C p
artic
ipat
ion
degr
ee fr
om u
pstr
eam
Domestic share of total CO2 embodied emisssions in gross exports
2009
: 30 Million Ton Foreign CO2 emissionsembodied in gross exports
by the way of non-energy final goods trade by the way of non-energy intermedaite goods tradeby the way of energy-related final goods tradeby the way of energy-related intermediate goods trade
: Embodied foreign carbon intensity 1.00 Kt/Million US$: Embodied foreign carbon intensity 0.75 Kt/Million US$: Embodied foreign carbon intensity 0.60 Kt/Million US$
VS1 share
47
3.5 Consumption-based versus production-based CO2 emissions and emissions transfer through different GVC routes
As shown by Peters et al. (2011), most developed countries (as Annex B countries in the Kyoto
Protocol) have increased their consumption-based CO2 emissions faster than their territorial
emissions. The net emissions transfer via international trade from developing to developed
countries increased very rapidly and exceeds the Kyoto Protocol emissions reduction. Expanding
on Peters et al. (2011) (use the forward industrial-linkage-based decomposition method
summarized by Figure 1), we not only estimate the consumption-based and production-based
emissions and their evolution from 1995 to 2009 for both Annex B and Non-Annex B country
groups, but also further investigate how the international transfer of emissions occurs through
various GVC routes with different environmental costs (carbon intensities).
Figure 12 shows that production-based CO2 emissions for the Annex B country group
have increased slightly in the period 1995-2009. Emission exports for satisfying foreign final
demands is the main driver of this increase, since territory emissions for fulfilling domestic final
demands have shown a slight decrease in the same period. Consumption-based emissions for the
Annex B country group experienced an increase due to increasing emissions imports (foreign
emissions induced by Annex B countries). Looking at the structure of Annex B countries’
increasing emissions trade by different GVC routes, we find that trade in intermediate goods is
the main contributor to growth for both exports and imports, with little change in trade through
final goods except for a slight increasing trend for imports. Compared to the Annex B countries,
the Non-Annex B country group shows large increases in both domestic emissions and emissions
trade. The production-based emissions for the Non-Annex B group in 2003 exceeded the Annex
B group’s peak level emissions (2007); Non-Annex B group’s territory emissions for its domestic
final demands in 2009 were close to the level of production-based emissions for Annex B groups.
The Non-Annex B country group also imports more emissions and has been at the same level as
the Annex B group’s emissions exports.
With the information about carbon intensity (the dark the color, the higher the emission
intensity with higher environment cost for per unit GDP; emissions in KT / GDP in million
US$ at 1995 constant prices) along different GVC routes, the major facts observed from Figure
12 can be summarized as follows:
48
49
Figure 12 Consumption-based vs. production-based CO2 emissions and emissions transfer through different GVC routes (1995-2009)
0
4000
8000
12000
CO2 emissions for trade partner's finaldemand through intermedaite goods trade bythe way of third countriesCO2 emissions for trade partner's finaldemand through intermedaite goods trade
CO2 emissions for trade partner's finaldemand through final goods trade
CO2 emissions for domestic final demandthrough international trade
CO2 emissions for domestic final demandwithout through international trade
20072009
1995.
. .2003.
Non-Annex B countries(developing economies)
Carbon intensity1.60 K Ton/Million US$1.60 K Ton/Million US$1.70 K Ton/Million US$1.20 K Ton/Million US$
0.45 K Ton/Million US$0.45 K Ton/Million US$0.30 K Ton/Million US$0.25 K Ton/Million US$
-4000 Unit: Million Ton
Emiss
ion
Expo
rtTe
rrito
ryEm
issio
nsfo
r ful
fillin
g do
mes
tic
final
dem
and
Emiss
ion
Impo
rt
Consumption-basedEmissions
Production-basedEmissions
19952007
2009. . .
Annex B countries(developed economies)
50
51
1. The environmental cost for generating one unit GDP in domestic production networks is
lower than that through international trade for both developed and developing countries.
One of the main drivers is the carbon leakage through international trade due to
differences in environmental regulation level across countries. Another driver is the
increasing fragmentation of production, which requires more international transportation
shipment (high-carbon intensity sector) across multiple borders multiple times.
2. The environmental cost for generating one unit GDP shows a decreasing trend for both
Annex B and Non-Annex B counties from 1995 to 2009. However, the carbon intensity
for Non-Annex B countries in 2009 is still higher than that for Annex B countries’ 1995
level. In addition, the decrease on carbon intensity14 in developing economies cannot
offset the increased emissions from rapid economic and population growth. This clearly
implies that helping more developing countries set carbon emission peak as China did in
2014 is more urgent than decades ago.
3. The increasing sophistication in cross country production sharing also give an impetus to
emissions transfer, since more cross-border CO2 emissions transfer arises through
intermediate goods trade via third countries.
3.6 The hidden environment cost of China’s comparative advantage in manufacturing
exports
As discussed in section 2, different measures of emission defined in this paper provide different
tools to quantify embodied CO2 emissions trades from different perspectives15. To provide a
better understanding of the differences between these measures and their economic and policy
implications, we use both the forward and the backward industrial-linkage-based domestic
emission measure to compute China’s Released Comparative Advantage (RCA16) as an example.
14 For detailed empirical results on carbon intensity at the bilateral level by different energy types along GVCs, one can refer to Figure B3 in Appendix. 15 Table B5 in Appendix B reports bilateral embodied emissions trade of Electrical and Optical Equipment (WIOD sector 14) between China and Japan in 2009 by different measures defined in section 2. It is a numerical example to illustrate the analytical relations among various emission trade measures we discussed in table 1b in real world data. 16 The RCA indicator used in the paper follows the additional RCA measure proposed by Hoen and Oosterhaven (2006). This type of indicator ranks from -1 to +1, with a symmetric distribution that centers on a stable mean of zero, independent of the sector classifications used.
52
The traditional RCA indicator (Balassa 1966) is based on gross exports. As pointed by
Wang et al. (2013), the traditional RCA ignores both domestic production sharing and
international production sharing. A conceptually correct measure of comparative advantage
needs to exclude foreign-originated value added and pure double counted terms in gross exports
but include indirect exports of a sector’s value added through other sectors of the exporting
country. When a country uses imported intermediate goods intensively to produce its exports,
Koopman et al. (2014) show that RCA based on gross exports can be very misleading and
suggested a way to remove the distortion of double counting by focusing on domestic value-
added in exports. We follow the same idea here to measure a country’s RCA by using both value-
added exports and CO2 emissions exports. As mentioned earlier, according to the forward
industrial-linkage-based decomposition, a country’s value-added or CO2 emissions exports at the
sector level represent how much of this country’s specific sector’s value-added or CO2 emissions
embodied in all downstream countries’ and sectors’ gross output is finally consumed in foreign
countries. According to the backward industrial-linkage-based decomposition, a country’s value-
added or CO2 emissions exports at the sector level measures how much this country’s value-
added or CO2 emissions in all upstream production stages are embodied in a specific product that
is finally consumed in foreign countries.
The upper penal of Figure 13 shows China’s forward industrial linage based RCA by
sector ranking for both value-added and CO2 emissions exports. For value-added exports,
Electrical and Optical Equipment (ICT, WIOD sector 14), Textiles and Textile Products (WIOD
sector 4) and Agriculture, Hunting, Forestry and Fishing (WIOD sector 1) show the highest RCA
since all these sectors generate more value-added for fulfilling foreign countries’ final demand
through global value chains directly and indirectly. However, for CO2 emissions exports, these
Chinese products are relative cleaner, only Electricity, Gas and Water Supply (sector 17) shows
an extremely high RCA. This implies that energy sector emits large amounts of CO2 emissions
embodied in China’s various manufacturing exports to satisfy foreign final demands, which are
not show up in traditional trade statistics since there is a negligible amount of Chinese electricity
exported directly.
The bottom penal of Figure 13 shows the backward industrial linkage based RCA estimates for
China. Clearly, the RCA for value-added export is normally consistent to that for CO2 emissions
53
export at the sector level. The production of Chinese textile and ICT exports is much more
carbon intensive due to its upstream sectors (such as electricity, metal, glass production) are
more carbon intensive than most developed countries. We see that from the perspective of a
producer, the production process of these Chinese products has a low-carbon intensity (forward),
but from the viewpoint of foreign user, they have a high-carbon intensity since relatively large
shares of CO2 emissions are generated in their upstream sectors (backward). This implies that
both downstream-driven and upstream-driven RCA indicators have their own roles in helping
better understanding the fact that China’s comparative advantage in many manufacturing sectors
in the world market are highly related to high-carbon inputs coming from their upstream sectors,
which have little direct exports in the traditional trade statistics, but is embodied in other Chinese
manufacturing products and in fact indirectly exports to the world market extensively.
Figure 13 Backward vs. forward industrial linkage based RCA for both value-added exports and CO2 emissions exports (2009)
54
-0.1
-0.05
0
0.05
0.1
0.15
0.2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
Back
war
d in
dust
rial
link
age
base
d R
CA
Textiles
ICT
-0.1
-0.05
0
0.05
0.1
0.15
0.2
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35
Forw
ard
indu
stri
al li
nkag
e ba
sed
RC
A
Value-added exportCO2 emission export
Agriculture ICT
Electricity
Textiles
55
4. Concluding remarks
The rise of global value chains has dramatically changed the nature and structure of international
trade in recent decades. There is particularly strong growth in intermediate goods and services
that may cross borders multiple times before the delivery of final products. This makes it
difficult to understand “who produces value for whom” in a fragmented production system,
compared to the relatively simple situation in the Ricardian era where exports were mainly final
goods. The increasing complexity of GVCs has produced challenges for economic and
environment policy as well as international governance. Therefore, it is important to understand
to what extent GVCs impact on both value creation and emissions generation for trade and
environment policies.
This paper unifies and extends existing emissions trade related measures, quantify their
relationships, and further combines them with trade in value-added and GVC-based measures in
recent literature into one consistent accounting framework, in which both value added and
emissions can be systematically traced at country, bilateral, and sector levels through various
GVC routes. In principle, when new countries or years are added to the WIOD database, or an
alternative inter-country input-output table becomes available, our accounting framework can be
applied as well. So the accounting framework developed in this paper is not inherently tied to the
WIOD database and can be a stand-alone tool. It provides a useful analytical method for both
trade and environment economists as well as policy makers to study the impact of production
fragmentation and emergence of GVCs on the environment. We show that conventional analysis
on carbon emission transfer, shared responsibilities and the environment cost of a country’s
comparative advantages can all benefit from applying such new analytical tool developed in this
paper.
Better and detailed information that combine environment cost and economic benefit in
each production stages and trade routes along GVCs provide useful insights regarding to the role
of each specific trade route in emission transfer and scientific evidence for concrete, targeted
incentive mechanism and an integrated trade and greenhouse gas emission reduction policy
design. We leave further analysis of the full decomposition results (it takes up 20 gigabytes of
space) and link it to policy design for our future research agenda.
56
57
References
Andrew, R.M. and Peters, G.P. (2013). A multi-region input-output table based on the Global Trade Analysis Project Database (GTAP-MRIO). Economic Systems Research 25 (1): 99-121.
Antràs, P. and Chor, D. (2013). Organizing the global value chain. Econometrica, 81(6): 2127-2204.
Arto, I. and Dietzenbacher, E. (2014). Drivers of the growth in global greenhouse gas emissions. Environmental Science & Technology 48(10): 5388−5394.
Atkinson, G., Hamilton, K., Ruta, G., Mensbrugghe, D. van der (2011). Trade in ‘virtual carbon’: Empirical results and implications for policy. Global Environmental Change 21(2): 563-574.
Atkinson, G., (2013). EU corporate action as a driver for global emissions abatement: A structural analysis of EU international supply chain carbon dioxide emissions. Global Environmental Change 23(3): 1795-1806.
Balassa, B. (1965). Trade Liberalization and Revealed Comparative Advantage. the Manchester School, 33 (2): 99-123.
Baldwin, R. (2012). WTO 2.0: Global governance of supply-chain trade. CEPR Policy Insight 64.
Baldwin, R. and Robert-Nicoud, F. (2014). Trade-in-goods and trade-in-tasks: An integrating framework. Journal of International Economics 92(1): 51-62.
Bastianoni, S., Pulselli, F.M., Tiezzi, E. (2004). The problem of assigning responsibility for greenhouse gas emissions. Ecological Economics 49: 253-257.
Cadarso, M.A., Lopez, L.A., Comez, N., Tobarra, M.A. (2012). International trade and shared environmental responsibility by sector. An application to the Spanish economy. Ecological Economics 83: 221-235.
58
Davis, S.J. and Caldeira, K. (2010). Consumption-based accounting of CO2 emissions. PNAS 107(12): 5687-5692.
Davis, S.J., Peters, G.P., and Caldeira, K. (2011). The supply chain of CO2 Emissions. PNAS 108 (45): 18554-18559.
de Haan, M. and Keuning, S. J. (1996). Taking the environment into account: The NAMEA approach. Review of Income and Wealth 42: 131-148.
de Haan, M. and Keuning, S. J. (2001). The NAMEA as validation instrument for environmental macroeconomics. Integrated Assessment 2: 79-87.
Ferrarini, B. and D. Hummels, eds. (2014) Asia and Global Production Networks –Implications for Trade, Incomes and economic Vulnerability. Edward Elgar.
Feng, J.J. (2003) Allocating the responsibility of CO2 over-emissions from the perspectives of benefit principle and ecological deficit. Ecological Economics 46: 121-141.
Grossman, G.M., Rossi-Hansberg, E. (2008). Trading tasks: a simple theory of offshoring. The American Economic Review 98(5): 1978-1997.
Hertwich, E.G. and Peters, G.P. (2009). Carbon footprint of nations: A global, trade-linked analysis. Environmental Science and Technology 43 (16): 6414-6420.
Hoekstra, A. and Mekonnen, M. (2012). The water footprint of humanity, PANS, 109(9), 3232-3237.
Hoekstra, A. and Wiedmann, T. (2014). Humanity’s unsustainable environmental footprint. Science 344(6188): 1114-1117.
Hoen, A.R. and Oosteraven, J. (2006). On the measurement of comparative advantage. The Annals of Regional Science 40: 677-691.
Hummels, D., Ishii, J., Yi, K.M. (2001). The nature and growth of vertical specialization in world trade. Journal of International Economics 54 (1): 75-96.
59
Johnson, R. and Noguera, G. (2012). Accounting for Intermediates: Production Sharing and Trade in Value added. Journal of International Economics 86: 224-236.
Kanemoto, K., Lenzen, M., Peters, G.P., Moran, D., Geschke, A. (2012). Frameworks for comparing emissions associated with production, consumption, and international trade. Environmental Science and Technology 46: 172-179.
Koopman, R., Wang, Z., Wei, S.J. (2014). Tracing value-added and double counting in gross exports. American Economic Review 104(2): 459-494.
Lenzen, M., Moran, D, Kanemoto, K., Foran, B., Lobefaro, L., Geschke, A. (2012), International trade drives biodiversity threats in developing nations. Nature 486: 109-112.
Lenzen, M., Murray, J., Sack, F., Wiedmann, T. (2007). Shared producer and consumer responsibility – theory and practice. Ecological Economics 61: 27-42.
Leontief, W. (1936). Quantitative input and output relations in the economic system of the United States. The Review of Economic and Statistics 18: 105-25.
Lin, J., Pan, D., Davis, S.J., Zhang, Q., He, K., Wang, C., Streets, D.G., Wuebbles, D.J., Guan, D. (2014). China’s international trade and air pollution in the United States. PNAS 111(5): 1736-1741.
Meng, B., Xue, J., Feng, K., Guan, D., Fu, X. (2013). China’s inter-regional spillover of carbon emissions and domestic supply chains. Energy Policy 61: 1305-1321.
OECD, WTO and UNCTAD (2013) “Implications of Global Value Chains for Trade, Investment Development and Jobs” prepared by OECD, WTO and UNCTAD for the G20 Leaders Summit, St. Petersburg, Russian Federation September.
Pedersen, G.O. and de Haan, M. (2006). The system of environmental and economic accounts-2003 and the economic relevance of physical flow accounting. Journal of Industrial Ecology 10: 19-42.
60
Peters, G.P., and Hertwich, E.G. (2006). The importance of imports for household environmental impacts. Journal of Industrial Ecology 10 (3): 89-109.
Peters, G.P. (2008). From production-based to consumption-based national emission inventories. Ecological Economics 65: 13-23.
Peters, G.P. and Hertwich, E.G. (2008). CO2 embodied in international trade with implications for global climate policy. Environmental Science and Technology, 42 (5): 1401-1407.
Peters, G.P., Minx, J.C., Weber, C.L., Edenhofer, O. (2011). Growth in emission transfers via international trade from 1990 to 2008. PANS 108(21): 8903-8908.
Peters, G.P., Davis, S.J., Andrew, R. (2012). A synthesis of carbon in international trade. Biogeosciences 9: 3247–3276.
Rodrigues, J., Domingos, T., Giljum, S., Schneider, F. (2006). Designing an indicator of environmental responsibility. Ecological Economics 59: 256-266. Timmer, M.P., Dietzenbacher, E., Los, B., Stehrer, R., de Vries,G.J. (2014a). The World Input-Output Database (WIOD): Contents, concepts and applications. GGDC Research Memorandum 144. Groningen Growth and Development Centre. Timmer, M.P., Erumban, A.A., Los, B., Stehrer, R., De Vries, G.J., (2014b). Slicing up global value chains. Journal of Economic Perspectives 28(2): 99-118.
Timmer, M. P., Los, B., Stehrer, R. and de Vries, G. J. (2013). Fragmentation, incomes and jobs: an analysis of European competitiveness. Economic Policy 28(76): 613–661.
Tukker, A. and Dietzenbacher, E. (2013). Global multiregional input–output frameworks: An introduction and outlook. Economic Systems Research 25(1): 1-19.
Wang, Z., Wei, S.J., Zhu, K. (2013). Quantifying international production sharing at the bilateral and sector levels. NBER Working Paper 19677.
Weinzettel, J., Hertwich, E.G., Peters, G.P., Steen-Olsen, K., Galli, A. (2013). Affluence drives the global displacement of land use. Global Environmental Change 23(2): 433-438.
61
Wiedmanna, T.O., Schandlb, H. Lenzen, M., Moranc, D., Suh, S., West, J., Kanemoto, K. (2013). The material footprint of nations. PANS.
Yi, K.M. (2010). Can multistage production explain the Home Bias in trade? The American Economic Review 100(1): 364-393.
62
Appendix A Detailed mathematical proofs17
A.1 Step by step proof of Equation (10) in the main text
Write 1)( −−= ssss AIL , then the last term of equation (9) in the main text can be written as
)(* ∑∑≠≠
+=G
sr
rsrG
sr
srsssss XAYLEL (A1)
Using the gross output rX decomposition equation
∑ ∑=G
t
G
u
turtr YBX ,
*sE can be expressed as
∑∑ ∑
∑∑ ∑∑ ∑∑∑∑
∑∑ ∑∑
≠≠ ≠
≠≠ ≠≠ ≠≠≠≠
≠≠
++
+++=
+=
G
sr
ssrssrtsG
sr
G
st
rtsr
G
tsu
tuG
sr
G
st
rtsrttG
sr
G
st
rtsrG
st
stG
sr
rssrG
sr
sr
G
u
tuG
sr
G
t
rtsrG
sr
srs
YBAYBA
YBAYBAYBAY
YBAYE
,
*
(A2)
Rearranging gives
ssG
st
tsstrsG
st
G
sr
trst
G
sr
urG
st
G
rsu
tustrrG
st
G
sr
trstG
sr
srG
st
tsstG
sr
srs
YBAYBA
YBAYBAYBAYE
∑∑ ∑
∑∑ ∑∑ ∑∑∑∑
≠≠ ≠
≠≠ ≠≠ ≠≠≠≠
++
+++=,
*
(A3)
Inserting equation (A3) into (A1) gives
ssG
st
tsstssrsG
st
G
sr
trstss
G
sr
trG
su
G
rst
utsussrrG
st
G
sr
trstssG
sr
srG
st
tsstsssssss
YBALYBAL
YBALYBALYBALLEL
∑∑ ∑
∑∑ ∑∑ ∑∑∑
≠≠ ≠
≠≠ ≠≠ ≠≠≠
++
++
+=
,
*
(A4)
17 We acknowledge Dr. Kunfu Zhu’s help on related mathematical derivations.
63
Using the properties of inverse matrices, we can obtain the identity
−−−
−−−−−−
=
=
−−−
−−−−−−
GGGG
G
G
GGGG
G
G
GGGG
G
G
GGGG
G
G
AIAA
AAIAAAAI
BBB
BBBBBB
I
II
BBB
BBBBBB
AIAA
AAIAAAAI
21
22221
11211
21
22221
11211
21
22221
11211
21
22221
11211
00
0000
(A5)
From (A5) we obtain
0)( =−− ∑≠
G
st
trstsrss BABAI (A6)
∑∑≠≠
−−==−−G
sr
rssrssssG
sr
rssrssss ABAIBIBABAI )()( (A7)
From equations (A6) and (A7), we can obtain following relationships between diagonal global block
inverse matrices and local inverse matrices:
∑≠
+=G
st
tsstssssss BALLB , ∑≠
=G
st
trstsssr BALB ,
∑≠
=G
sr
rtsrssst BALB , ssG
sr
rssrG
st
tsstss LABBAL ∑∑≠≠
=
Inserting these four equations into (A4) gives
ssssG
sr
rssrG
sr
rssrG
rst
rtG
sr
srG
sr
rrsrG
sr
srsssss YLABYBYBYBYBEL ∑∑∑∑∑∑≠≠≠≠≠≠
++++=,
* (A8)
which is exactly the same as equation (10) in the main text. We can further show that
∑ ∑∑∑ ∑∑∑≠≠≠ ≠≠≠
=+=+G
st
G
r
rstrstssG
st
tsstrsG
st
G
sr
trstssssG
sr
rssrG
sr
rssr YBAYBAYBAYLABYB (A9)
64
A.2 Step by step proofs of Equations (18), (19) and (20) in the main text
As equation (1) in the main text shows, the gross exports of country s to country r can be decomposed
into two parts: final goods exports and intermediate goods exports,
rsrsrsr XAYE += (A10)
As illustrated in section 2.3 in the main text (equation 14), emission embodied in final goods exports
can be easily decomposed into domestic and foreign emission by directly applying Leontief inverse.
However, the decomposition of emission embodied in intermediate goods trade flows is more complex. It
cannot be achieved by simply multiplying the Leontief inverse with gross intermediate exports because
the latter has to be solved from the MRIO models first for any given level of final demand. Wang et al.
(2013) provide a method to overcome this endogeneity issue by expressing all intermediate trade flows as
different countries’ final demands according to where the goods or services are absorbed. Following their
method, the gross output of country r can be decomposed into the following components according to
where it is finally absorbed (obtained from equation (12) in the main text by pick-up country r only):
∑∑∑∑∑∑∑∑∑
∑∑∑ ∑∑∑∑
∑∑ ∑∑∑ ∑
≠≠≠≠≠ ≠≠≠ ≠
≠ ≠≠ ≠≠ ≠≠
≠≠ ≠
+++++
+++=
++==
G
rst
stG
sr
rssrG
sr
rsG
sr
ssrstsG
sr
G
st
rtG
trsu
tuG
sr
G
rst
rt
trG
sr
G
rst
rtG
sr
G
st
rtrrttG
sr
G
rst
rtrrG
sr
rr
G
st
strsG
rst
G
tsu
turtG
t
rtrrG
t
G
u
turtr
YBYBYBYBYB
YBYBYBYB
YBYBYBYBX
,,,,
,,
, ,
(A11)
Inserting equation (A11) into the last term of equation (A10), the gross intermediate exports of
country s to country r can be fully decomposed according to where they are absorbed:
∑∑∑∑∑ ∑∑∑ ∑
∑ ∑∑ ∑∑ ∑∑
≠≠≠≠≠ ≠≠≠ ≠
≠ ≠≠ ≠≠ ≠≠
+++++
+++=
G
rst
stG
sr
rssrsrG
sr
rssrG
sr
ssrssrtsG
sr
G
st
rtsrG
trsu
tuG
sr
G
rst
rtsr
trG
sr
G
rst
rtsrG
sr
G
st
rtrrsrttG
sr
G
rst
rtsrrrG
sr
rrsrrsr
YBAYBAYBAYBAYBA
YBAYBAYBAYBAXA
,,,,
,,
(A12)
This decomposition is intuitively illustrated by figure A1.
After laying out the idea of how bilateral gross intermediate trade flows are decomposed, we
provide a detailed step by step proof in a 3-country setting to simplify notation and make the materials
accessible to more readers. Inserting equations (A10) and (A12 ) into the left hand of equation (19) in the
65
main text, which defines domestic emissions embodied in gross exports from country s to country r based
on forward industrial linkages, we obtain
[ ][ ][ ]ssrssrsstsrtsrssrsrrsrss
strssrssttrtsrssrtrrsrss
srrssrsstrrtsrssrrrrsrsssrss
srsssr
YBALYBALYBAVLF
YBALYBALYBALF
YBALYBALYBALFYLF
ELFFEEG
s
s
ss
s
+++
+++
+++=
=
∧
∧
∧∧
∧
_
(A13)
66
Figure A1. Accounting for gross bilateral intermediate trade flows between country s and country r
Source: improved from Wang, Wei and Zhu (2014) Learning about global value chains by looking beyond official trade data: Part 1. http://www.voxeu.org/article/learning-about-global-value-chains-looking-beyond-official-trade-data-part-1
Re-exports to s
*srssr YBA
rrrrsr YBAtt
G
srt rtsr YBA ∑ ≠ ,ssrssr YBA
*rrrsr YBA
Direct
importing Intermediate exports
*kG
k rksrrsr YBAXA ∑=
Third
Exporting
*, t
G
rst rtsr YBA ∑ ≠
Final goods consumed in s
Final goods consumed in t
Final goods consumed in r
Re-exports to
Intermediate trade
Final trade flows
rtrrsr YBArsrrsr YBA
trrtsr YBAtsrtsr YBA
strssr YBA
Domestic final good flows
67
Re-exports to S
The image part with relationship ID rId278 was not found in the file.
The image part with relationship ID rId278 was not found in the file.The image part with relationship ID rId278 was not found in the file.
68
The 1st term, srssYLF s∧
, represents emissions generated by each industry of country s embodied in its final
goods exports to country r. The 2nd-4th terms (the 1st bracket) are emissions generated by each industry of
country s embodied in its intermediate exports to country r that are driven by final demand in country r.
The 5th-7th terms (the 2nd bracket) are emissions generated by each industry of country s embodied in its
intermediate exports to country r that are driven by final demand in third countries (t). The 8th-10th terms
(the 3rd bracket) are emissions generated by each industry of country s embodied in its intermediate
exports to country r that ultimately return and are driven by final demand in country s.
Based on equation (17) in the main text, EEX_Fsr, emission exports from country s to country r
based on forward industrial linkage in a three country world can be expressed as
[ ] [ ][ ] [ ][ ] [ ]trttstsstrrtsrssrrtrstssrrrrsrss
srtsstsssrrssrsssrsstrttstsstrrtsrss
rrtrstssrrrrsrsssrsssssrss
trstrrsrsrsssr
YBALYBALFYBALYBALF
YBALYBALFYLFYBALYBALF
YBALYBALFYLBYLF
YBFYBFYBFFEEX
ss
sss
ss
sss
++++
++=++
++−+=
++=
∧∧
∧∧∧
∧∧
∧∧∧
)(
_
(A14)
Rearranging equation (A14) gives
[ ][ ]srtsstsstrttstssrrtrstss
srrssrsstrrtsrssrrrrsrsssrsssr
YBALYBALYBALF
YBALYBALYBALFYLFFEEXs
ss
+++
+++=∧
∧∧
_
(A15)
Therefore,
[ ][ ][ ]srtsstsstrttstssrrtrstss
strssrssttrtsrssrtrrsrss
ssrssrsstsrtsrssrsrrsrss
trstrrsrsrsssrsssrsr
YBALYBALYBALF
YBALYBALYBALF
YBALYBALYBALF
YBFYBFYBFELFFEEXFEEG
s
s
s
ssss
++−
+++
++=
++−=−
∧
∧
∧
∧∧∧∧
__
(A16)
The 1st bracket of equation (A16) is emissions by industry embodied in the intermediate exports of
country s to country r that are ultimately returned to satisfy final demand at home, which is the same as
equation (18) in the main text in a three country world. We call it REE_Fsr:
∑∧
∧∧∧
=
++=G
u
usrusrss
ssrssrsstsrtsrssrsrrsrsssr
YBALF
YBALFYBALFYBALFFREE
s
sss_
(A17)
69
The 2nd bracket in equation (A16) represents emissions by industry embodied in the intermediate
exports from country s to country r that are driven by final demand in the third country (t). The 3rd bracket
in equation (A16) represents emissions by industry embodied in the intermediate exports of country s to
the third country (t) that are driven by final demand in country r. It is easy to understand that the 2nd and
the 3rd brackets in equation (A16) are not equal to each other except very special cases. Therefore, EEG_F
or srss ELF s∧
based on forward linkage does not equal EEX_F + REE_F at bilateral and bilateral sector
level.
However, summing up equation (A16) over all trade partners (i.e., countries r and t in the three
country world), the terms in the 2nd bracket and the terms in the 3rd bracket will equal each other and
cancel out:
[ ] [ ][ ]
[ ][ ]
[ ]stsr
strssrssttrtsrssrtrrsrss
srtsstsstrttstssrrtrstssst
srtsstsstrttstssrrtrstss
strssrssttrtsrssrtrrsrsssr
ststsssrsrss
FREEFREEYBALYBALYBALF
YBALYBALYBALFFREE
YBALYBALYBALF
YBALYBALYBALFFREE
FEEXELFFEEXELF
s
s
s
s
ss
__
_
_
__
+=
++−
++++
++−
+++=
−+−
∧
∧
∧
∧
∧∧
(A18)
Rearranging equation (A18) gives
[ ] [ ]ststsrsr
stsssrssstsr
FREEFEEXFREEFEEX
ELFELFFEEGFEEG ss
______
+++=
+=+∧∧
(A19)
Therefore, EEG_F or srss ELF s∧
based on forward linkage are equal to EEX_F + REE_F at the
country/sector and country aggregate levels. This proves that equation (20) in the main text holds.
A.3 Step by step proofs of Equations (25), (26) and (27) in the main text
Inserting equations (A10) and (A12 ) into the left hand side of equation (25) in the main text, which
defines domestic emissions embodied in gross exports from country s to country r based on backward
industrial linkages, we obtain the following equations for the three country world.
70
)(#)()(#)(
)(#)(#)(#)(_
ssrssrtsrtstrsrrsrTsss
strssrrtrrsrttrtsrTsss
srrssrtrrtsrrrrrsrTssssrTssssrTssssr
YBAYBAYBALFYBAYBAYBALF
YBAYBAYBALFYLFELFBEEG
+++
+++
+++==(A20)
This shows that EEG_Bsr can be decomposed into four parts: emissions embodied in final goods exports,
emissions embodied in intermediate goods that are used to satisfy final demand in the direct importing
country r, emissions embodied in intermediate exports returned to the exporting country s, and re-
exported to third countries t. Emissions in these terms include emissions generated not only by the
exporting sectors but also by other domestic sectors that contribute to the production of a particular
sector’s gross exports.
Based on equation (23) in the main text, EEX_Bsr can be expressed as
)(#)()(#)()(#)()(#)(#)(_
trttstTsssrrtrstTsss
trrtsrTsssrrrrsrTssssrTssssr
YBALFYBALFYBALFYBALFYBFBEEX
++
++=
(A21)
where
srTtsstssssrTrssrssssrTsss
srTsssssssrTssssrTsss
YBALFYBALFYLFYLFBFYLFYBF
#)(#)(#)(#)(#)(#)(
++=
−+=
(A22)
Inserting equation (A22) into equation (A21) we obtain
)(#)()(#)()(#)()(#)(
#)(#)(#)(_
trrtsrTsssrrtrstTsss
trttstTsssrrrrsrTsss
srTtsstssssrTrssrssssrTssssr
YBALFYBALFYBALFYBALF
YBALFYBALFYLFBEEX
++
++
++=
(A23)
Therefore
]#)(#)[(]#)()(#)[(
)(#)()(#)(
_#)(
srTrssrssssrrssrTsss
srTtsstssstrttstrrtrstTsss
strssrrtrrsrttrtsrTsss
ssrssrtsrtsrrsrrsrTsss
srsrTsss
YBALFYBALFYBALFYBAYBALF
YBAYBAYBALFYBAYBAYBALF
BEEXELF
−+
++−
+++
++=
−
(A24)
71
The first term of equation (A24) represents the amount of emissions embodied in the sectoral
exports from country s to country r that finally return home, and is exactly the same as equation (26) in
the main text in a three country world:
)(#)()(#)()(#)(_ ssrssrTssstsrtsrTsssrsrrsrTssssr YBALFYBALFYBALFBREE ++=
(A25)
The second term of equation (A24) represents emissions in the sectoral intermediate exports of
country s to country r which are then re-exported to other countries (both countries r and s) to produce
final products that are consumed in the third country t. The third term of equation (A24) represents
emissions in the gross intermediate exports of country s to third country t to produce final product exports
to country r or produce intermediate products exports to countries r or s for production of final goods and
services consumed in country r. As we will show later, srTrssrssssrrssrTsss YBALFYBALF #)(#)( = at the
bilateral aggregate level but not at the bilateral/sector level.
Therefore
0]#)()(#)[(]#)(#)[(
)(#)(___
≠++−
−+
++=−−
srTtsstssstrttstrrtrstTsss
srTrssrssssrrssrTsss
strssrrtrrsrttrtsrTssssrsrsr
YBALFYBAYBALFYBALFYBALF
YBAYBAYBALFBREEBEEXBEEG
(A26)
It is obvious that the positive and negative terms in equation (A26) are not equal to each other except in
very special cases. This indicates that EEG_Bsr and (EEX_Bsr +REE_ Bsr ) cannot be equal each to other
at the bilateral/sector level in general. At the bilateral aggregate level, summing (A26) over sectors, we
obtain
0)()()(#)(
)(#)(___
≠++−
++=
++−
++=−−
srtsstsstrttstssrrtrstsss
strssrssrtrrsrssttrtsrsss
srtssttrttstrrtrstTsss
strssrrtrrsrttrtsrTssssrsrsr
YBALYBALYBALFYBALYBALYBALFYBAYBAYBALFu
YBAYBAYBALFuBuREEBuEEXBuEEG
(A27)
The two terms in equation (A27) are still not equal each other in general. Therefore, the sum of
srBuEEX _ and srBuREE _ does not equal srBuEEG _ at the bilateral aggregate level.
Summing up equation (A27) over all trading partners r and t, the positive and negative terms will
cancel out:
72
0)()()()(
)____(__
=++−
+++
++−
++=
−−−−+
strssrssrtrrsrssttrtsrsss
srtsstsstrttstssrrtrstsss
srtsstsstrttstssrrtrstsss
strssrssrtrrsrssttrtsrsss
ststsrsrstsr
YBALYBALYBALFYBALYBALYBALFYBALYBALYBALFYBALYBALYBALF
BREEBEEXBREEBEEXuBuEEGBuEEG
(A28)
Therefore, equation (27) in the main text holds.
∑∑∑≠≠≠
=+=G
rs
srsssG
sr
srsrG
sr
sr ELFBuREEBuEEXBuEEG )__(_
In a two-sector case,
[ ]
[ ]
≠
−
−=
−
=
+
+−
++++++
++
=
−
=
−
∑∑∑∑
∑∑∑∑
∑∑∑
∑∑∑
∑∑∑
∑∑∑
∑∑∑∑
∑∑∑∑
00
#
#
#
#
#)(#)(
22
2
1
2
111
2
2
2
2
11
2
2
2
222
2
1
2
1
22
222
11
222
22
2
2
2
22
1
2
1
2
1
2
22
222
22
11
1
22
221
22
11
22222121222122111121
22212121122121111111
222121
212111
2
1
2221
1211
2221
1211
2221
121121
2
1
2221
1211
2212
1211
2221
121121
srrsj
j
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srrsk
k
srjk
j
ssij
i
si
srrsk
k
srjk
j
ssij
i
si
srk
k
rsjk
j
srj
i
ssi
si
srk
k
rsjk
j
srj
i
ssi
si
sr
sr
rsj
j
srij
i
ssi
srsj
j
srij
i
ssi
s
rsj
j
srij
i
ssi
srsj
j
srij
i
ssi
s
srrssrsrrssrsrrssrsrrssr
srrssrsrrssrsrrssrsrrssr
ssssss
ssssss
sr
sr
rsrs
rsrs
srsr
srsr
ssss
ssssss
sr
sr
rsrs
rsrs
srsr
srsr
ssss
ssssss
srTrssrssssrrssrTsss
ybalfybalf
ybalfybalf
ybalf
ybalf
ybalf
ybalf
yy
balfbalf
balfbalf
ybaybaybaybaybaybaybayba
lflflflf
yy
bbbb
aaaa
llll
ff
yy
bbbb
aaaa
llll
ff
YBALFYBALF
(A29)
However,
73
[ ] 011
#)(#)(
22
2
1
2
111
2
2
2
2
11
2
2
2
222
2
1
2
1
=
−
−=
−
∑∑∑∑
∑∑∑∑srrs
jj
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srrsj
j
srj
i
ssi
si
srTrssrssssrrssrTsss
ybalfybalf
ybalfybalf
YBALFuYBALFu
(A30)
Both elements in the last term in (A29) are not equal to zero in general. However, after aggregating
over sectors, the two elements will cancel each other, as shown in equation (A30). Therefore, summing up
equation (A26) over all trading partners r and t, but not over sectors, the positive and negative terms will
not cancel out, as in equation (A27). This means∑≠
G
sr
srBEEG _ is also not equal to the sum of
∑≠
G
sr
srBEEX _ and∑≠
G
sr
srBREE _ at the country-sector level.
74
Appendix B Additional empirical results based on WIOD
B1 Who emits CO2 emissions for whom
Table B1 shows how much CO2 emissions are induced by different sources of final demand through
different routes of supply chains in both 1995 and 2009 for selected large countries . From the upper part
of Table B1 we see that China’s total production-based CO2 emissions experienced the largest increase
(128%) from 2,723,066 kt in 1995 to 6,213,385 kt followed by India (108%) and the rest of the world
(RoW, 37%)18. For all developed countries, their production-based CO2 emissions decreased, especially
for Germany which had the largest decline of 12%.
Total production-based CO2 emissions can be decomposed into 5 parts (referring to Figure 1)
according to final demand in different segments of global market they satisfied. The structure and
changing pattern among these five final demand markets between 1995 and 2009 are shown in the middle
and bottom parts of Table B1. Obviously, for all selected countries and for both years, the CO2 emissions
generated by the domestic production of goods and services that sell directly in the domestic market
(EH_F) account for the majority of the total emissions, especially for countries with relatively large
economic size. This is not surprising because most large countries’ production is mainly for domestic use.
The interesting thing is that the share of the remaining 4 segments of these final demand markets show a
very different pattern across countries. For example, in both 1995 and 2009, the share of China’s CO2
emissions generated by its production of final goods exports (EEX_F1) is the largest when compared to
the other selected countries. This implies that China’s participation in GVCs is mainly through providing
final goods exports and, naturally, relatively more CO2 emissions are generated by this route. In contrast,
Russia’s CO2 emissions generated by foreign final demand are mainly from providing intermediate goods
exports (EEX_F2 + EEX_F3). This phenomenon clearly illustrates that a country’s production-based CO2
emissions depend not only on the energy efficiency of its production technology, but also on its position
and participation in GVCs. Both Germany and UK have a large portion of their production-based CO2
emissions that are generated by the production of exports to meet foreign final demand, as China does,
18The RoW here is not the rest of the selected countries shown in Table 1; it’s the original country group of the RoW used in WIOD regarded as a group of all the other developing countries not covered by WIOD.
75
but with a much higher portion of such emissions generated by the production of intermediate exports.
When looking at the changing pattern of the shares between 1995 and 2009 (the bottom right part of
Table B1), for most countries except India, EH_F decreased, while other parts normally increased. This
reflects the fact that most countries have been involved in GVCs and more of their emissions production
is for satisfying final demands in foreign countries. In particular, the increase in the share for EEX_F2 is
about 61% (from 9.1% to 14.7%) for China, and 63% (from 13.0% to 21.3%) for Germany. Since both
countries have been the main supply hub of intermediate manufacturing goods in international trade, a
relatively large portion of CO2 emissions are naturally generated by this route. The share for EEX_F3
(emissions generated by the production of intermediates that re-exported to third countries) is lower than
EEX_F1 and EEX_F2, while its rate of change for all countries is positive and very large. This clearly
reflects the increasing complexity of GVCs, since more intermediate goods and services cross national
borders more than once and are re-exported to third countries for further processing in the global
production networks. In addition, the share for REE_F also experienced a dramatic increase for all
selected developing countries, such as China (592%), India (294%) and the RoW (123%), although the
absolute level of this share is extremely low. This implies that the final goods imported by China tend to
embody more emissions generated by its own intermediate goods exports given its increasing presence in
international production networks.
76
Table B1 CO2 emissions induced by different segments of global final demand (forward industrial-linkage-based decomposition, corresponding to Figure 1)
CO2 Emissions(KT)
EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum
CHN 2,126,639 3,196 301,045 249,125 43,061 2,723,066 4,191,734 50,471 891,922 913,035 166,223 6,213,385 IND 607,263 165 39,284 65,961 8,154 720,827 1,266,226 1,356 95,723 116,290 22,214 1,501,809 JPN 874,562 3,068 43,965 90,214 12,458 1,024,267 753,151 3,223 47,700 124,446 25,217 953,737 USA 3,869,470 38,148 142,285 262,327 29,954 4,342,184 3,719,713 29,436 136,290 264,124 38,152 4,187,715 GBR 316,770 2,228 42,859 75,658 13,517 451,032 285,484 2,015 40,381 79,426 14,991 422,297 DEU 542,851 7,014 61,628 94,494 18,717 724,704 383,503 7,692 81,929 135,490 27,695 636,309 RUS 974,488 3,278 48,382 326,921 59,269 1,412,338 926,130 3,731 34,581 360,665 85,379 1,410,486 RoW 2,626,249 30,223 218,217 442,696 59,812 3,377,197 3,341,296 92,569 292,962 784,936 129,232 4,640,995
Share(%)
EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum
CHN 78.1% 0.1% 11.1% 9.1% 1.6% 100.0% 67.5% 0.8% 14.4% 14.7% 2.7% 100.0%IND 84.2% 0.0% 5.4% 9.2% 1.1% 100.0% 84.3% 0.1% 6.4% 7.7% 1.5% 100.0%JPN 85.4% 0.3% 4.3% 8.8% 1.2% 100.0% 79.0% 0.3% 5.0% 13.0% 2.6% 100.0%USA 89.1% 0.9% 3.3% 6.0% 0.7% 100.0% 88.8% 0.7% 3.3% 6.3% 0.9% 100.0%GBR 70.2% 0.5% 9.5% 16.8% 3.0% 100.0% 67.6% 0.5% 9.6% 18.8% 3.5% 100.0%DEU 74.9% 1.0% 8.5% 13.0% 2.6% 100.0% 60.3% 1.2% 12.9% 21.3% 4.4% 100.0%RUS 69.0% 0.2% 3.4% 23.1% 4.2% 100.0% 65.7% 0.3% 2.5% 25.6% 6.1% 100.0%RoW 77.8% 0.9% 6.5% 13.1% 1.8% 100.0% 72.0% 2.0% 6.3% 16.9% 2.8% 100.0%
EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum
CHN 97% 1479% 196% 266% 286% 128% -14% 592% 30% 61% 69%IND 109% 722% 144% 76% 172% 108% 0% 294% 17% -15% 31%JPN -14% 5% 8% 38% 102% -7% -8% 13% 17% 48% 117%USA -4% -23% -4% 1% 27% -4% 0% -20% -1% 4% 32%GBR -10% -10% -6% 5% 11% -6% -4% -3% 1% 12% 18%DEU -29% 10% 33% 43% 48% -12% -20% 25% 51% 63% 69%RUS -5% 14% -29% 10% 44% 0% -5% 14% -28% 10% 44%RoW 27% 206% 34% 77% 116% 37% -7% 123% -2% 29% 57%
Change rate between1995 and 2009
1995 2009
Change rate of CO2 emisions between 1995 and 2009 Change rate of shares between 1995 and 2009
77
B2 CO2 emissions generated in domestic and foreign segments of global supply chains
As shown in Figure 2, a country’s CO2 emissions can also be traced along global supply chains in terms
of different types of energy source by using the backward industrial-linkage-based decomposition
technique. Table B2 shows the decomposition results at the national level (sector aggregation) for
selected countries in 1995 and 2009. In absolute terms, in 1995, the US’s production of final products, no
matter whether they are used domestically or internationally, generates the largest amount of CO2
emissions (4,423,852 kt). The US is followed by the RoW (3,382,085 kt) and China (2,513,050 kt). This
depends both on a country’s economic size and on its energy efficiency. In 2009, the situation changed
dramatically: with a 125% increase compared to 1995, China becomes the largest emitter, followed by the
RoW, the US and India. When looking at the share (the middle part of Table B3), we can see that CO2
emissions generated in domestic segments of global supply chains accounts for the majority of total
induced CO2 emissions for all selected countries. This can be easily understood since, for most countries,
their upstream supply chains are mainly located at home. However, the difference of the share across
countries is still significant. For example, more than 20% of CO2 emissions from Japan’s, the UK and
Germany’s production of final products are generated in foreign segments of global supply chains in
1995. This clearly reflects at least two facts: one is that these countries’ supply chains need more foreign
intermediate inputs for producing final products, and the other is that much higher CO2 emission intensity
is located in foreign segments of their global supply chains than for the other selected developing
countries.
The structure of energy use for producing final products in global supply chains varies across
countries. China’s and India’s CO2 emissions generated in their domestic supply chains are mainly from
the use of coal (76.0% and 64.1% respectively in 1995). This depends not only on their relatively rich
endowment of coal, but also on the higher CO2 emission intensity in production processes using coal.
This can also be indirectly confirmed by the fact that most of the CO2 emissions generated in the foreign
segment of Japan’s supply chains were from coal in 2009, since most of its foreign upstream industries
are located in China, which provides intermediate products mainly by using coal-based energy.
When looking at the pattern of structure changes between 1995 and 2009 (the bottom part of
Table B2), some important features emerge. 1) For all selected countries, the share of CO2 emissions
78
generated in the domestic segment of their global supply chains declined, especially for China (-6.4%),
England (-7.1%), Germany (-7.9%), and the RoW (-8.7%). On the other hand, the share of their foreign
segments increased dramatically, especially for China (186%). Since countries tend to use more
intermediate imports to make final goods, given the reduction in international trade costs, naturally more
CO2 emissions are generated in foreign segments of supply chains. 2) The share of coal, petroleum, and
other energy-based CO2 emissions generated in the domestic segment decreased, while natural gas and
waste-based CO2 emissions increased between 1995 and 2009. This reflects the fact that more countries
are shifting to the usage of relatively low carbon intensity energy in the domestic part of their final goods
production. Japan is the only exception, its coal-based CO2 emissions in domestic segment increased
32.0 % from 1995 to 2009. This is mainly because Japan’s energy efficiency is higher even if using coal
to generate energy rather than thermal power generation; at the same time, it is cheaper to import coal
from neighboring countries, like China which is a coal-rich country. 3) For almost all emission sources,
their shares of CO2 emissions in the foreign segment for all selected countries increased significantly
between 1995 and 2009. In this regard, China’s change is the most remarkable. This is mainly because
China has been both the largest final goods assembler and a producer which also needs to import more
components and intermediate inputs produced by foreign countries.
79
Table B2 CO2 emissions to produce a final goods and services in global supply chains (backward industrial-linkage-based decomposition, corresponding to Figure 2)
1995
CO2 emissions(Kt)
Coal Petroleum Gas Waste Other Subtotal Coal Petroleum Gas Waste Other Subtotal
CHN 1,911,062 293,157 38,157 - 187,373 2,429,749 23,052 31,061 18,937 386 9,865 83,301 2,513,050 IND 439,230 139,432 24,262 - 43,743 646,667 11,451 12,235 9,829 174 5,027 38,716 685,383 JPN 236,609 484,494 125,142 2,703 71,315 920,263 95,738 96,867 53,407 664 29,841 276,517 1,196,780 USA 1,641,832 1,421,481 731,322 35,302 198,759 4,028,696 120,695 139,960 85,996 1,332 47,173 395,156 4,423,852 GBR 139,308 116,119 71,457 1,191 32,567 360,642 37,565 41,270 24,354 786 10,758 114,733 475,375 DEU 307,303 197,880 87,580 8,777 6,097 607,637 84,962 73,667 62,218 2,475 27,492 250,814 858,451 RUS 260,885 215,568 451,172 9,283 87,242 1,024,150 7,602 7,172 4,209 178 3,297 22,458 1,046,608 RoW 614,637 1,393,462 639,832 3,633 210,533 2,862,097 162,491 232,758 77,264 2,158 45,317 519,988 3,382,085 Share (%) Coal Petroleum Gas Waste Other Subtotal Coal Petroleum Gas Waste Other Subtotal TotalCHN 76.0% 11.7% 1.5% 0.0% 7.5% 96.7% 0.9% 1.2% 0.8% 0.0% 0.4% 3.3% 100.0%IND 64.1% 20.3% 3.5% 0.0% 6.4% 94.4% 1.7% 1.8% 1.4% 0.0% 0.7% 5.6% 100.0%JPN 19.8% 40.5% 10.5% 0.2% 6.0% 76.9% 8.0% 8.1% 4.5% 0.1% 2.5% 23.1% 100.0%USA 37.1% 32.1% 16.5% 0.8% 4.5% 91.1% 2.7% 3.2% 1.9% 0.0% 1.1% 8.9% 100.0%GBR 29.3% 24.4% 15.0% 0.3% 6.9% 75.9% 7.9% 8.7% 5.1% 0.2% 2.3% 24.1% 100.0%DEU 35.8% 23.1% 10.2% 1.0% 0.7% 70.8% 9.9% 8.6% 7.2% 0.3% 3.2% 29.2% 100.0%RUS 24.9% 20.6% 43.1% 0.9% 8.3% 97.9% 0.7% 0.7% 0.4% 0.0% 0.3% 2.1% 100.0%RoW 18.2% 41.2% 18.9% 0.1% 6.2% 84.6% 4.8% 6.9% 2.3% 0.1% 1.3% 15.4% 100.0%
2009CO2 emissions (Kt) Coal Petroleum Gas Waste Other Subtotal Coal Petroleum Gas Waste Other SubtotalCHN 4,098,564 552,773 142,473 0 326,088 5,119,898 161,716 170,108 146,806 3,421 54,990 537,041 5,656,939 125%IND 952,788 244,857 79,460 0 85,728 1,362,833 57,762 36,723 32,685 510 13,875 141,555 1,504,388 119%JPN 274,427 306,539 168,896 7,356 45,322 802,540 101,801 73,519 53,700 749 19,254 249,023 1,051,563 -12%USA 1,632,018 1,259,978 798,603 53,355 126,083 3,870,037 238,903 160,596 136,688 2,075 55,471 593,733 4,463,770 1%GBR 89,744 85,842 101,247 3,575 46,391 326,799 51,785 41,930 31,504 1,254 10,389 136,862 463,661 -2%DEU 214,441 146,990 85,506 21,330 278 468,545 98,039 67,708 57,925 2,050 24,767 250,489 719,034 -16%RUS 197,522 174,079 468,240 12,910 109,339 962,090 15,567 9,588 5,938 277 3,671 35,041 997,131 -5%RoW 761,424 1,644,039 1,048,100 6,930 230,144 3,690,637 455,449 395,188 155,364 6,249 72,088 1,084,338 4,774,975 41%Share (%) Coal Petroleum Gas Waste Other Subtotal Coal Petroleum Gas Waste Other Subtotal TotalCHN 72.5% 9.8% 2.5% 0.0% 5.8% 90.5% 2.9% 3.0% 2.6% 0.1% 1.0% 9.5% 100.0%IND 63.3% 16.3% 5.3% 0.0% 5.7% 90.6% 3.8% 2.4% 2.2% 0.0% 0.9% 9.4% 100.0%JPN 26.1% 29.2% 16.1% 0.7% 4.3% 76.3% 9.7% 7.0% 5.1% 0.1% 1.8% 23.7% 100.0%USA 36.6% 28.2% 17.9% 1.2% 2.8% 86.7% 5.4% 3.6% 3.1% 0.0% 1.2% 13.3% 100.0%GBR 19.4% 18.5% 21.8% 0.8% 10.0% 70.5% 11.2% 9.0% 6.8% 0.3% 2.2% 29.5% 100.0%DEU 29.8% 20.4% 11.9% 3.0% 0.0% 65.2% 13.6% 9.4% 8.1% 0.3% 3.4% 34.8% 100.0%RUS 19.8% 17.5% 47.0% 1.3% 11.0% 96.5% 1.6% 1.0% 0.6% 0.0% 0.4% 3.5% 100.0%RoW 15.9% 34.4% 21.9% 0.1% 4.8% 77.3% 9.5% 8.3% 3.3% 0.1% 1.5% 22.7% 100.0%Change rate of the share
between 1995 and 2009 (%) Coal Petroleum Gas Waste Other Subtotal Coal Petroleum Gas Waste Other Subtotal Total
CHN -4.7% -16.2% 65.9% -22.7% -6.4% 211.6% 143.3% 244.4% 293.7% 147.6% 186.4% 0.0%IND -1.2% -20.0% 49.2% -10.7% -4.0% 129.8% 36.7% 51.5% 33.5% 25.7% 66.6% 0.0%JPN 32.0% -28.0% 53.6% 209.7% -27.7% -0.7% 21.0% -13.6% 14.4% 28.4% -26.6% 2.5% 0.0%USA -1.5% -12.2% 8.2% 49.8% -37.1% -4.8% 96.2% 13.7% 57.5% 54.4% 16.5% 48.9% 0.0%GBR -34.0% -24.2% 45.3% 207.8% 46.0% -7.1% 41.3% 4.2% 32.6% 63.6% -1.0% 22.3% 0.0%DEU -16.7% -11.3% 16.6% 190.1% -94.6% -7.9% 37.8% 9.7% 11.2% -1.1% 7.6% 19.2% 0.0%RUS -20.5% -15.2% 8.9% 46.0% 31.5% -1.4% 114.9% 40.3% 48.1% 63.3% 16.9% 63.8% 0.0%RoW -12.3% -16.4% 16.0% 35.1% -22.6% -8.7% 98.5% 20.3% 42.4% 105.1% 12.7% 47.7% 0.0%
Change rate
between
1995 and
2009
CO2 emissions generated by domestic segment of GVC CO2 emissions generated by foreign segment of GVCTotal
CO2 emissions generated by domestic segment of GVC CO2 emissions generated by foreign segment of GVC
Total
80
B3 CO2 emissions induced by the production of gross exports for selected countries
As shown in Figure 3, when applying the backward industrial-linkage-based decomposition technique, it
can identify who emits CO2 emissions for whom to what extent in the production of gross exports. Table
B3 reports the decomposition results for selected countries at the national level for both 1995 and 2009. In
absolute terms, the RoW’s gross exports induce the largest amount of CO2 emissions (869,561 kt) in
1995 followed by China (717,838 kt) and the US (531,191 kt). The total CO2 emissions can be separated
into domestic and foreign parts. The majority of induced CO2 emissions in producing exports were from
the domestic side for all selected countries. However, if a country, in producing exports, has a relatively
large part of the upstream production process outside its territory, the share of foreign CO2 emissions
could be large, as for Germany (33%), England (24%) and Japan (20%). Both the domestic part and the
foreign part can be further divided into 4 parts, each based on different supply chain routes and types of
final consumer. Obviously, in 1995, 97% of CO2 emissions embodied in China’s gross exports is from the
domestic side, in which 49% is for fulfilling final demand of trading partners who directly import goods
from China; 35% is for fulfilling China’s trading partners’ demands for intermediate inputs in their
production of domestically consumed goods and services; 13% is for fulfilling third countries’ final
demands by providing intermediate goods to China’s trading partners for their production of exports to
third countries; just 1% is for fulfilling China’s own final demand by re-importing what has been
exported. For most countries, except China, their domestic CO2 emissions embodied in gross exports
come mainly through trade in intermediate goods (parts 2, 3, 4). For Part 4, the figure for the US is larger
than all other countries. This is mainly because the US re-imports a relatively large part of its own
intermediate goods that have first been exported to global supply chains. For the foreign CO2 emissions in
producing gross exports, Germany shows the largest figure, in which parts 7 and 8 account for 17% and
15%, respectively. This indicates that 17% of the total CO2 emissions embodied in Germany’s gross
exports is from third countries which export intermediate goods to Germany for Germany’s further
production of final goods for export to its trading partners. On the other hand, 15% of the total CO2
emissions embodied in Germany’s gross exports is from third countries that export intermediate goods to
Germany, which uses these goods to produce further intermediate goods and exports to its trading
partners for making domestically consumed final goods and services. Part 5 shows the CO2 emissions
induced in Germany’s trading partner countries that provide intermediate goods to Germany for its
production of final goods which are finally consumed in its trading partner countries. Part 6 shows the
CO2 emissions induced in Germany’s trading partners which provide intermediate goods to Germany for
further processing into intermediate exports, which are imported by Germany’s trading partners for
81
producing domestically used final goods and services. Together parts 5 and 6 account for just 1%, since
this kind of feedback effect in international production networks is normally small.
In order to investigate the structural changes of gross-export-based CO2 emissions between 1995
and 2009 across different routes, we calculate the rate of change for both the absolute CO2 emissions
figure and the corresponding share and show the results in the bottom two parts of Table B3. We see the
following three features. 1) The induced CO2 emissions in gross exports for all developing countries, such
as China (262%), India (128%), and the RoW (85%), experienced a more rapid increase than developed
countries. Given the decreasing CO2 intensity, both for developing countries and developed countries
from 1995 to 2009, the most important driving factor for this change should be the rapid increase of gross
exports produced by developing countries. For England and the USA, there are only 1% and 5%
increases, respectively. Japan and Germany also experienced 37% and 48% increases, respectively.
Although both of them have been service oriented economies, they still play an important role as two
large trade hubs of intermediate goods in global supply chains. 2) When looking at the change of share,
we see that the share of domestic CO2 emissions in producing exports decreased for all countries, while
the share of foreign CO2 emissions increased for most countries, except England. This indirectly reflects
the fact that most countries are getting to use more intermediate imports to produce their exports. As a
result, relatively more CO2 emissions are induced internationally rather than domestically in producing
exports. 3) Looking at the changing pattern for each part, we see that parts 3, 7 and 8 have a relatively
large absolute share and also show a positive change of their shares between 1995 and 2009. Therefore,
these parts can be considered the main leading factors that cause both the increase in the absolute
emissions and the share of total gross-export-based CO2 emissions for all countries. All these three parts
are related to the third country effects in our decomposition. This implies that the increasing complexity
of global supply chains is often associated with a corresponding increase of CO2 emissions.
82
Table B3 CO2 emissions in the production of gross exports (backward industrial-linkage-based decomposition, corresponding to Figure 3)
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN 301,045 214,501 77,685 3,196 596,427 1,241 940 12,392 6,839 21,411 617,838
IND 39,284 58,469 15,646 165 113,563 211 335 2,117 2,537 5,200 118,763
JPN 43,965 78,316 24,356 3,068 149,705 1,933 3,015 14,999 18,493 38,439 188,144
USA 142,285 228,543 63,738 38,148 472,714 3,176 4,034 25,195 26,072 58,477 531,191
GBR 42,859 61,174 28,001 2,228 134,262 1,784 1,973 20,562 17,855 42,174 176,436
DEU 61,628 76,173 37,038 7,014 181,853 2,924 2,586 45,228 40,108 90,846 272,700
RUS 48,382 260,126 126,064 3,278 437,850 85 286 993 3,679 5,043 442,893
RoW 218,217 382,331 120,177 30,223 750,948 5,530 5,760 50,908 56,416 118,613 869,561
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN 49% 35% 13% 1% 97% 0% 0% 2% 1% 3% 100%IND 33% 49% 13% 0% 96% 0% 0% 2% 2% 4% 100%JPN 23% 42% 13% 2% 80% 1% 2% 8% 10% 20% 100%USA 27% 43% 12% 7% 89% 1% 1% 5% 5% 11% 100%GBR 24% 35% 16% 1% 76% 1% 1% 12% 10% 24% 100%DEU 23% 28% 14% 3% 67% 1% 1% 17% 15% 33% 100%RUS 11% 59% 28% 1% 99% 0% 0% 0% 1% 1% 100%RoW 25% 44% 14% 3% 86% 1% 1% 6% 6% 14% 100%
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN 891,922 764,257 315,000 50,471 2,021,650 16,375 15,473 109,535 75,942 217,325 2,238,975
IND 95,723 92,687 45,817 1,356 235,583 2,634 2,029 21,564 9,298 35,524 271,107
JPN 47,700 98,451 51,212 3,223 200,586 3,276 7,268 19,022 27,921 57,487 258,073
USA 136,290 220,410 81,866 29,436 468,002 5,376 7,886 36,705 39,913 89,880 557,881
GBR 40,381 62,046 32,372 2,015 136,814 1,592 2,249 19,409 18,977 42,227 179,040
DEU 81,929 105,433 57,752 7,692 252,806 5,599 6,615 75,059 63,183 150,456 403,262
RUS 34,581 254,843 191,202 3,731 484,356 143 591 919 4,147 5,800 490,157
RoW 292,962 658,916 255,252 92,569 1,299,699 8,670 18,993 120,711 157,417 305,791 1,605,490
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN 40% 34% 14% 2% 90% 1% 1% 5% 3% 10% 100%IND 35% 34% 17% 1% 87% 1% 1% 8% 3% 13% 100%JPN 18% 38% 20% 1% 78% 1% 3% 7% 11% 22% 100%USA 24% 40% 15% 5% 84% 1% 1% 7% 7% 16% 100%GBR 23% 35% 18% 1% 76% 1% 1% 11% 11% 24% 100%DEU 20% 26% 14% 2% 63% 1% 2% 19% 16% 37% 100%RUS 7% 52% 39% 1% 99% 0% 0% 0% 1% 1% 100%RoW 18% 41% 16% 6% 81% 1% 1% 8% 10% 19% 100%
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN 196% 256% 305% 1479% 239% 1220% 1547% 784% 1010% 915% 262%IND 144% 59% 193% 722% 107% 1151% 506% 919% 266% 583% 128%JPN 8% 26% 110% 5% 34% 69% 141% 27% 51% 50% 37%USA -4% -4% 28% -23% -1% 69% 95% 46% 53% 54% 5%GBR -6% 1% 16% -10% 2% -11% 14% -6% 6% 0% 1%DEU 33% 38% 56% 10% 39% 91% 156% 66% 58% 66% 48%RUS -29% -2% 52% 14% 11% 69% 106% -7% 13% 15% 11%RoW 34% 72% 112% 206% 73% 57% 230% 137% 179% 158% 85%
part 1 part 2 part 3 part 4 subtotal part 5 part 6 part 7 part 8 subtotalCHN -18% -2% 12% 336% -6% 264% 354% 144% 206% 180%IND 7% -31% 28% 260% -9% 448% 165% 346% 61% 199%JPN -21% -8% 53% -23% -2% 24% 76% -8% 10% 9%USA -9% -8% 22% -27% -6% 61% 86% 39% 46% 46%GBR -7% 0% 14% -11% 0% -12% 12% -7% 5% -1%DEU -10% -6% 5% -26% -6% 29% 73% 12% 7% 12%RUS -35% -11% 37% 3% 0% 53% 87% -16% 2% 4%RoW -27% -7% 15% 66% -6% -15% 79% 28% 51% 40%
Chage rate ofshare (%)
Domestic CO2 emissions in producing exports Foreign CO2 emissions in supplying imported inputsTotal
Between 1995 and 2009
2009
Chage rate ofCO2 emisions (%)
Domestic CO2 emissions in producing exports Foreign CO2 emissions in supplying imported inputsTotal
Share(%)
Total
CO2 emissions(KT)
Share(%)
Domestic CO2 emissions in producing exports Foreign CO2 emissions in supplying imported inputsTotal
Domestic CO2 emissions in producing exports Foreign CO2 emissions in supplying imported inputs
Domestic CO2 emissions in producing exports Foreign CO2 emissions in producing exportsTotal
1995CO2 emissions
(KT)
Domestic CO2 emissions in producing exports Foreign CO2 emissions in producing exportsTotal
83
B4 The potential environmental cost of value-added trade
As discussed in the main text, in our decomposition frameworks, both value-added and embodied
emissions can be traced simultaneously. When dividing the induced value added by induced CO2
emissions, the potential environmental cost can be easily obtained. As an example, we apply this idea to
the forward industrial-linkage-based decomposition (Figure 1) to show the relationship between trade in
value added and trade in CO2 emissions.
Table B4 The potential environmental cost of trade in value added (using forward industrial-linkage-based decomposition)
CO2 emissions/value-added(KT/Million US$)
EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum
CHN 3.6 4.6 3.9 4.6 4.3 3.7IND 1.8 3.5 2.5 3.4 3.1 1.9JPN 0.2 0.4 0.3 0.4 0.3 0.2USA 0.6 0.7 0.7 0.7 0.7 0.6GBR 0.4 0.6 0.5 0.6 0.6 0.4DEU 0.3 0.4 0.3 0.4 0.4 0.3RUS 3.9 5.9 4.2 6.0 6.4 4.4RoW 1.0 1.5 1.4 1.4 1.5 1.1
CO2 emissions/value-added(KT/Million US$)
EH_F REE_F EEX_F1 EEX_F2 EEX_F3 Sum
CHN 2.1 2.8 2.3 2.7 2.6 2.2IND 1.6 2.7 1.8 2.2 2.3 1.6JPN 0.2 0.4 0.3 0.4 0.3 0.2USA 0.4 0.5 0.5 0.5 0.5 0.4GBR 0.2 0.4 0.4 0.4 0.4 0.3DEU 0.2 0.3 0.2 0.3 0.3 0.2RUS 2.4 4.3 3.0 4.1 4.1 2.8RoW 0.8 1.0 1.1 1.0 1.1 0.8
Change rate (%) EH_F REE_F EEX_F1 EEX_F2 EEX_F3 SumCHN -41% -40% -40% -42% -40% -40%IND -13% -24% -28% -35% -23% -16%JPN -13% -4% 0% 0% 2% -8%USA -31% -27% -23% -29% -29% -31%GBR -33% -36% -9% -33% -34% -31%DEU -32% -24% -22% -24% -27% -26%RUS -39% -27% -29% -31% -35% -36%RoW -25% -34% -24% -29% -27% -24%
1995
2009
between 1995 and 2009
84
The main results are shown in Table B4. In general, the environmental cost for producing
domestic value added without international trade (referring to EH_F) for all countries is lower than that of
producing domestic value added through international trade. This implies that the value-added gain by
international trade may be through a high-carbon process, which indirectly reflects the fact of carbon
leakage across countries due to trade. At the country level, Russia shows the highest environmental cost
(4.4 kt/million US$) followed by China (3.7 kt/million US$) in 1995, which are, respectively 18.5 and
22.0, times more costly than Japan (0.2 kt/million US$). In 2009, for all countries, a cost decrease can be
observed, especially for China (-40%) and Russia (-36%). Energy efficiency changes and emissions-
related regulation conducted both domestically and internationally can be considered as the main driving
factors of this decline. However, the situation regarding carbon leakage shows no significant change,
since the environmental cost for getting value added by international trade is still higher than that for pure
domestic production in 2009.
B5 CO2 emissions generated in the foreign segment of global supply chains by specific products
The backward industrial-linkage-based decomposition technique can help us trace the CO2 emissions in
supply chains at the detailed sector level for production of a specific final good in a particular country. As
an example, Figure B1 shows the foreign sectors with the largest CO2 emissions (top 30 out of 1435
sectors across all WIOD countries) in China’s and Germany’s Transportation Equipment supply chains
for both 1995 and 2009. The major features can be summarized as follows. 1) The most intensive emitters
of upstream countries in both countries’ Transportation Equipment supply chains are from their
neighboring countries. This is not surprising, since parts and components for producing cars follow the
so-called just-in-time production system and trade costs across countries is one of the most important
factors that affect the choice of production locations. It is, therefore, reasonable to build supply chains
regionally rather than globally. 2) For both China and Germany, the most intensive foreign sector emitters
in their Transportation Equipment supply chains are sectors 17 (Electricity, Gas and Water Supply), 12
(Basic Metals and Fabricated Metal), 9 (Chemicals and Chemical Products), and 2 (Mining and
Quarrying). This depends on how close and strong the upstream sector links with the final product of
transportation equipment, as well as the intensity of the CO2 emissions arising from the production of
parts and components directly and indirectly in the relevant upstream sectors. 3) Dramatic changes occur
in the rankings of upstream countries and sectors during the 15 year sample period. This reflects the
evolution of competitiveness not only in the quality and price of an upstream country or sector’s
85
intermediate goods in supply chains, but also on their energy efficiency. 4) The foreign segments in
German car production are greener than those of China.
86
Figure B1 Foreign sectoral CO2 emissions (top 30 sectors) induced by a specific country's production of final goods (Transportation Equipment) in global supply chains
050010001500200025003000
RUS_Sector12RUS_Sector17RoW_Sector17RoW_Sector12JPN_Sector12KOR_Sector12USA_Sector17TWN_Sector17RoW_Sector9RoW_Sector10RoW_Sector2TWN_Sector12RoW_Sector8USA_Sector12KOR_Sector17TWN_Sector9AUS_Sector12JPN_Sector17KOR_Sector9USA_Sector9RUS_Sector2AUS_Sector17RUS_Sector9AUS_Sector2DEU_Sector17JPN_Sector9CAN_Sector2RoW_Sector14DEU_Sector12IND_Sector17
Foreign CO2 emissions induced by China's final goods production of transportation equipment, 1995
0 500 1000 1500 2000 2500 3000
RoW_Sector17RoW_Sector12
RoW_Sector9KOR_Sector17
RoW_Sector2RUS_Sector17RUS_Sector12JPN_Sector12
RoW_Sector10TWN_Sector17KOR_Sector12USA_Sector17JPN_Sector17RoW_Sector8IND_Sector17
DEU_Sector17RoW_Sector14
AUS_Sector2RoW_Sector16
RUS_Sector2AUS_Sector17TWN_Sector9
TWN_Sector12IDN_Sector2
CAN_Sector2IND_Sector2
RoW_Sector24JPN_Sector9
CHN_Sector6USA_Sector12
Foreign CO2 emissions induced by China's final goods production of transportation equipment, 2009
050010001500200025003000350040004500
RUS_Sector17RUS_Sector12RoW_Sector12RoW_Sector17POL_Sector12POL_Sector17CHN_Sector17RUS_Sector2BEL_Sector12CZE_Sector12RoW_Sector10RoW_Sector9CZE_Sector17USA_Sector17FRA_Sector12CHN_Sector12RUS_Sector23GBR_Sector12AUT_Sector12ITA_Sector12SVK_Sector12ITA_Sector17USA_Sector12ROM_Sector17NLD_Sector9GBR_Sector17NLD_Sector12RoW_Sector16RoW_Sector2CAN_Sector2
Foreign CO2 emissions induced by Germany's final goods production of
transportation equipment, 19950 500 1000 1500 2000 2500 3000 3500 4000 4500
CHN_Sector17RoW_Sector17RUS_Sector12RoW_Sector12CHN_Sector12RUS_Sector17POL_Sector17RoW_Sector9
RoW_Sector10TWN_Sector24
IND_Sector17AUT_Sector12USA_Sector17CZE_Sector12POL_Sector12CHN_Sector9
CZE_Sector17RUS_Sector2
CHN_Sector11RoW_Sector2
KOR_Sector17GBR_Sector12FRA_Sector12KOR_Sector12
CHN_Sector2SVK_Sector12NLD_Sector12ITA_Sector17RoW_Sector8
RUS_Sector23
Foreign CO2 emissions induced by Germany's final goods production of
transportation equipment, 2009
87
B6 Impacts of bilateral trade on CO2 emissions
In order to elucidate how bilateral trade flows between China and Non-Annex B* countries impact on the
global environment, the use of EEG_B measure should be a better choice. As we discussed in section 2,
EEG_B is a production side concept, only concern the amount of emission generated by the production of
a particular bilateral trade flow regardless where these traded products and services were consumed, so
the emissions embodied in intermediate exports but final return to the source country are included. Figure
B2 compares both the share of value-added and CO2 emission embodied in the bilateral trade between
China and the US with Non-Annex B* countries as a share of GDP or emissions embodied in global trade
respectively. It clearly shows that there are opposite trends for China-Non-Annex B* and US- Non-Annex
B* bilateral flows. The embodied CO2 emissions share for China-Non-Annex B* countries experiences
significant growth (from 5% to 19%), while the share of the US- Non-Annex B* countries has be in
decline (from 13% to 9%). More remarkable difference can be observed in the share of coal based
embodied CO2 emissions, which the share of China-Non-Annex B* countries increased from 10% to
29%, but the share of US- Non-Annex B* countries has decreased from 9% to 5% over the same period.
This clearly indicates that the bilateral trade flows between China and Non-Annex B* countries became
darker and darker over last two decades, increasingly became the major source of “carbon leakage” in the
global production and trading system.
Figure B2 Embodied CO2 emissions in bilateral trade between China (US) and Non-Annex B*
countries as a share of total embodied CO2 emissions in global trade
88
Note: Non-Annex B* excludes China.
Figure B3 The potential environmental costs at the bilateral level for different energy sources (2009, kt/million US$)
Coal-based CO2 Petroleum-based CO2
Natural gas-based CO2 Other source-based CO2
0%
10%
20%
30% Coal-based CO2 emissions, between China and Non-Annex B*Total CO2 emissions, between China and Non-Annex B*Value-added, between China and Non-Annex B*Coal-based CO2 emissions, between the US and Non-Annex B*Total CO2 emissions, between the US and Non-Annex B*Value-added, between the US and Non-Annex B*
89
90
Table B5 The relationships among different measures of embodied CO2 emissions and their applications
Level
ExampleEEX EEX_F EEX_B REE_F REE_B EEG_F EEG_B EEX_F+REE_F EEX_B+REE_B
Bilateral-sector(China→Japan,WIOD14)
38,634 867 39,206 31 1,395 880 39,427 898 40,601
Bilateral Aggregate (China→Japan) 147,839 147,022 147,022 4,645 4,645 152,256 152,256 151,667 151,667
Country-Sector(China→World,WIOD14)
557,698 12,463 557,698 428 19,804 12,891 574,614 12,891 577,502
Country Aggregate (China→World) 1,971,179 1,971,179 1,971,179 50,471 50,471 2,021,650 2,021,650 2,021,650 2,021,650
Indicators
91
Appendix C
C1 Sharing emission responsibility between producers and consumers along GVCs
A number of papers have discussed sharing responsibility between producers and consumers (Feng, 2003;
Bastianoni et al., 2004; Rodrigues et al., 2006; Lenzen et al., 2007; Peters, 2008; Cadarso et al., 2012).
However, two important problems remain unsolved. One is about how to correctly identify a country’s
pure self-responsibility of emissions along GVCs. Without a correct measure on this part, we even not
able to know the amount of emission should be shared among related parties. This problem has been
solved in our paper (see the first part in Equation 11). The pure self-responsibility of emissions is defined
as the emissions generated in production of domestic consumed final goods and services without through
any route of international trade (Part 1 in Figure 1). Another unsolved issue is about how to find an
objective weight to share responsibility between producers and consumers. Lenzen et al. (2007) proposes
to use value added as a weight, Cadarso et al., (2012) also follow this idea. However, there is an
endogeneity problem due to value-added production is not independent to the emission level. In order to
share responsibility more reasonably, we propose a new way to first measure the carbon leakage from
both producers and consumer’s perspectives based on the following hypothesis: if a country wants to
keep its current final demand level in an autarky world, its emissions are defined as the emissions that
this country just uses domestic production technology without importing any intermediate inputs to fulfill
the same level of final demand as international trade exists. Compared this autarky emissions with both
current production based and consumption based emissions, two indicators can be computed: production
based carbon leakage and consumption based carbon leakage. These two indicators can be considered the
carbon leakage that the country should take responsibility as a producer and a consumer respectively, thus
the weight of shared responsibility can be obtained (for definition in mathematical terms and algorithm,
refer to Appendix C2). Table 2 shows the results of shared emissions responsibility between producers
and consumers for 41 economies in 2009. In the extreme case, that all responsibility goes to producers,
China accounts for 29.8% followed by the RoW (19.2%), Russia (7.1%), the US (6.9%), Germany (3.7)
and Korea (3.3%),. If all responsibility goes to consumers, the RoW accounts for 22.8% followed by the
92
US (16.1%), China (7.9%), Germany (6.0%), and Japan (5.8%). Based on the shared responsibility we
proposed, China should take 22.3%, the RoW 17.7%, the US 11.8%, Russia 7.1%, and Germany 4.4%.
93
Table C1 Shared responsibility of CO2 emissions along GVCs by country in 2009
2009unit: Kt
Productionbased
emissions
Consumptionbased
emissions
AutarkyEmissions
Production-based
leakage
Consumption-basedleakage
Production-based
contributionto carbon
leakage bycountry
Consumption-based
contributionto carbon
leakage bycountry
Share ofResponsibility
as producer
Share ofResponsibilityas consumer
Self-responsibili
ty
Productionbased
emissionsshould to be
shared
Consumptionbased
emissionsshould to be
shared
Productionbased
responsibilityonly
Consumptionbased
responsibilityonly
SharedProduction
basedresponsibility
SharedConsumption
basedresponsibility
Finalresponsibility
by country
Finalresponsibility
by country
PP PC AECLP=
PP-AECLC=
PC-AECLPS CLCS
∅=CLPS/
(CLPS+CLCS)
1-∅=CLCS/
(CLPS+CLCS)SE PPT=PP-SE PCT=PC-SE
share of PPTby country
share of PCTby country
FSP FSC FS share
AUS 364,325 414,091 311,892 52,433 102,199 1.1% 2.1% 33.9% 66.1% 277,544 86,781 136,547 1.3% 2.0% 24,576 75,370 99,946 1.5%AUT 47,928 81,033 28,543 19,385 52,490 0.4% 1.1% 27.0% 73.0% 22,271 25,657 58,762 0.4% 0.9% 5,779 35,840 41,619 0.6%BEL 91,053 116,888 48,898 42,155 67,990 0.8% 1.4% 38.3% 61.7% 34,114 56,939 82,774 0.8% 1.2% 18,200 42,672 60,872 0.9%BGR 41,684 33,288 28,097 13,587 5,191 0.3% 0.1% 72.4% 27.6% 21,671 20,013 11,617 0.3% 0.2% 12,094 2,682 14,776 0.2%BRA 251,288 306,481 218,098 33,190 88,383 0.7% 1.8% 27.3% 72.7% 207,891 43,397 98,590 0.6% 1.5% 9,895 59,859 69,754 1.0%CAN 439,065 477,170 327,793 111,272 149,377 2.2% 3.0% 42.7% 57.3% 286,630 152,435 190,540 2.2% 2.8% 54,348 91,198 145,546 2.1%CHN 6,213,385 4,725,895 4,429,743 1,783,642 296,152 35.9% 6.0% 85.8% 14.2% 4,191,734 2,021,651 534,161 29.8% 7.9% 1,447,984 63,524 1,511,508 22.3%CYP 6,713 9,658 8,069 -1,356 1,589 0.0% 0.0% -582.0% 682.0% 5,524 1,189 4,134 0.0% 0.1% -5,779 23,546 17,767 0.3%CZE 96,801 88,508 64,332 32,469 24,176 0.7% 0.5% 57.3% 42.7% 53,311 43,490 35,197 0.6% 0.5% 20,819 12,546 33,365 0.5%DEU 636,309 793,786 453,403 182,906 340,383 3.7% 6.9% 35.0% 65.0% 383,503 252,806 410,283 3.7% 6.0% 73,798 222,885 296,682 4.4%DNK 78,220 58,506 26,864 51,356 31,642 1.0% 0.6% 61.9% 38.1% 22,227 55,993 36,279 0.8% 0.5% 28,935 11,551 40,486 0.6%ESP 230,728 313,198 188,144 42,584 125,054 0.9% 2.5% 25.4% 74.6% 162,766 67,962 150,432 1.0% 2.2% 14,418 93,721 108,139 1.6%EST 14,245 11,215 11,001 3,244 214 0.1% 0.0% 93.8% 6.2% 7,475 6,770 3,740 0.1% 0.1% 5,304 193 5,498 0.1%FIN 55,188 64,203 37,860 17,328 26,343 0.3% 0.5% 39.7% 60.3% 32,693 22,495 31,510 0.3% 0.5% 7,454 15,874 23,328 0.3%FRA 260,360 434,683 206,686 53,674 227,997 1.1% 4.6% 19.1% 80.9% 175,568 84,792 259,115 1.2% 3.8% 13,494 175,166 188,660 2.8%GBR 422,297 534,319 363,812 58,485 170,507 1.2% 3.4% 25.5% 74.5% 285,484 136,813 248,835 2.0% 3.7% 29,182 154,741 183,923 2.7%GRC 93,776 124,461 91,941 1,835 32,520 0.0% 0.7% 5.3% 94.7% 78,452 15,324 46,009 0.2% 0.7% 684 36,373 37,056 0.5%HUN 41,606 48,237 27,704 13,902 20,533 0.3% 0.4% 40.4% 59.6% 22,468 19,138 25,769 0.3% 0.4% 6,453 12,833 19,285 0.3%IDN 331,193 323,133 257,954 73,239 65,179 1.5% 1.3% 52.9% 47.1% 245,345 85,848 77,788 1.3% 1.1% 37,936 30,591 68,527 1.0%IND 1,501,808 1,458,813 1,330,284 171,524 128,529 3.5% 2.6% 57.2% 42.8% 1,266,226 235,582 192,587 3.5% 2.8% 112,471 68,897 181,368 2.7%IRL 27,569 47,161 20,326 7,243 26,835 0.1% 0.5% 21.3% 78.7% 15,954 11,615 31,207 0.2% 0.5% 2,062 20,524 22,586 0.3%ITA 329,336 459,195 268,285 61,051 190,910 1.2% 3.8% 24.2% 75.8% 237,923 91,413 221,272 1.3% 3.3% 18,499 140,021 158,519 2.3%JPN 953,737 1,147,716 800,104 153,633 347,612 3.1% 7.0% 30.7% 69.3% 753,151 200,586 394,565 3.0% 5.8% 51,346 228,525 279,871 4.1%KOR 532,878 469,954 341,918 190,960 128,036 3.8% 2.6% 59.9% 40.1% 310,646 222,232 159,308 3.3% 2.3% 111,105 53,402 164,507 2.4%LTU 11,527 16,407 7,929 3,598 8,478 0.1% 0.2% 29.8% 70.2% 5,908 5,619 10,499 0.1% 0.2% 1,398 6,156 7,554 0.1%LUX 3,039 7,169 1,461 1,578 5,708 0.0% 0.1% 21.7% 78.3% 1,197 1,842 5,972 0.0% 0.1% 333 3,907 4,240 0.1%LVA 7,181 9,910 5,233 1,948 4,677 0.0% 0.1% 29.4% 70.6% 4,399 2,782 5,511 0.0% 0.1% 683 3,249 3,932 0.1%MEX 351,280 384,635 303,997 47,283 80,638 1.0% 1.6% 37.0% 63.0% 278,366 72,914 106,269 1.1% 1.6% 22,508 55,947 78,455 1.2%MLT 2,514 3,448 2,330 184 1,118 0.0% 0.0% 14.1% 85.9% 1,533 981 1,915 0.0% 0.0% 116 1,373 1,489 0.0%NLD 166,194 179,325 86,684 79,510 92,641 1.6% 1.9% 46.2% 53.8% 69,900 96,294 109,425 1.4% 1.6% 37,143 49,179 86,322 1.3%POL 275,037 251,284 213,241 61,796 38,043 1.2% 0.8% 61.9% 38.1% 187,194 87,843 64,090 1.3% 0.9% 45,408 20,395 65,804 1.0%PRT 52,180 63,485 42,613 9,567 20,872 0.2% 0.4% 31.4% 68.6% 36,027 16,153 27,458 0.2% 0.4% 4,240 15,724 19,964 0.3%ROM 76,798 82,187 63,099 13,699 19,088 0.3% 0.4% 41.8% 58.2% 56,019 20,779 26,168 0.3% 0.4% 7,251 12,723 19,974 0.3%RUS 1,410,486 1,037,438 1,099,441 311,045 -62,003 6.3% -1.2% 124.9% -24.9% 926,130 484,356 111,308 7.1% 1.6% 505,227 -23,144 482,082 7.1%SVK 33,179 34,703 19,685 13,494 15,018 0.3% 0.3% 47.3% 52.7% 14,598 18,581 20,105 0.3% 0.3% 7,344 8,844 16,188 0.2%SVN 13,042 16,324 8,319 4,723 8,005 0.1% 0.2% 37.1% 62.9% 6,825 6,217 9,499 0.1% 0.1% 1,927 4,989 6,916 0.1%SWE 47,351 74,119 28,143 19,208 45,976 0.4% 0.9% 29.5% 70.5% 21,842 25,509 52,277 0.4% 0.8% 6,278 30,794 37,072 0.5%TUR 239,608 269,083 198,350 41,258 70,733 0.8% 1.4% 36.8% 63.2% 185,151 54,457 83,932 0.8% 1.2% 16,755 44,273 61,028 0.9%TWN 290,360 198,033 150,726 139,634 47,307 2.8% 1.0% 74.7% 25.3% 129,888 160,472 68,145 2.4% 1.0% 100,105 14,402 114,507 1.7%USA 4,187,715 4,812,099 3,958,044 229,671 854,055 4.6% 17.2% 21.2% 78.8% 3,719,713 468,002 1,092,386 6.9% 16.1% 82,833 718,974 801,807 11.8%RoW 4,640,995 4,888,737 3,821,591 819,404 1,067,146 16.5% 21.5% 43.4% 56.6% 3,341,296 1,299,699 1,547,441 19.2% 22.8% 471,458 731,038 1,202,496 17.7%Total 24 ,869,978 24,869,978 19,902,637 4,967,341 4,967,341 100.0% 100.0% 50.0% 50.0% 0 6,783,422 6,783,422 100.0% 100.0% 3,412,064 3,371,358 6,783,422 100.0%
94
95
C2 Method and algorithm for sharing emissions responsibility between producers and consumers along GVCs
In an autarky state, if a country wants to keep its current final demand level, its emissions are defined as
AE𝑠𝑠 = F𝑠𝑠𝐿𝐿𝑠𝑠𝑠𝑠 ∑ 𝑌𝑌𝑟𝑟𝑠𝑠𝑟𝑟 .
In other words, AEs represents the emission level that country s uses domestic production technique without any intermediate imports to produce goods and service for fulfilling the same final demand level as international trade exists. Compared this Autarky Emissions with both current production based and consumption based emission levels, it’s easy to get two indicators: production based carbon leakage and consumption based carbon leakage as shown below.
𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠 = 𝐶𝐶𝐶𝐶𝑠𝑠 − 𝐴𝐴𝐴𝐴𝑠𝑠,
𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠 = 𝐶𝐶𝐶𝐶𝑠𝑠 − 𝐴𝐴𝐴𝐴𝑠𝑠.
Clearly, 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠 can be considered the carbon leakage that country s should take responsibility as a producer; 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠 as the carbon leakage that country s should take responsibility as a consumer. Following this definition, the contribution level by country for both types of leakage can further be defined as
𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠 = 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠/∑ 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠𝑠𝑠 ,
𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠 = 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠/∑ 𝐶𝐶𝐿𝐿𝐶𝐶𝑠𝑠𝑠𝑠 .
The above contribution levels can be used to define producers’ and consumers’ responsibility shares (weights) respectively as
∅𝑠𝑠 = 𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠/(𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠 + 𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠),
(1 − ∅𝑠𝑠) = 𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠/(𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠 + 𝐶𝐶𝐿𝐿𝐶𝐶𝐶𝐶𝑠𝑠).
Removing the pure-self-responsibility based emissions (SE) from both production and consumption based emissions, the remained parts are the targets to be shared.
𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠 = 𝐶𝐶𝐶𝐶𝑠𝑠 − 𝐶𝐶𝐴𝐴𝑠𝑠,
𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠 = 𝐶𝐶𝐶𝐶𝑠𝑠 − 𝐶𝐶𝐴𝐴𝑠𝑠.
Following Peters (2008)’s idea, the shared responsibility is given as
96
𝐹𝐹𝐶𝐶 = ∑ 𝐹𝐹𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠 + ∑ 𝐹𝐹𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠
= ∑ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 + ∑ (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 .
It should be noted, that by definition,
∑ 𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠 = ∑ 𝐶𝐶𝐶𝐶𝑠𝑠𝑠𝑠 => ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 = ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 .
In the process of sharing responsibility with ∅𝑠𝑠, there is no guarantee in the first step that the shared responsibility
𝐹𝐹𝐶𝐶 = ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 𝑜𝑜𝑜𝑜 = ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 .
Here, we use the following iterative algorithm to share responsibility step by step.
𝐹𝐹𝐶𝐶𝑡𝑡=1 = ∑ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 + ∑ (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 .
𝐹𝐹𝐶𝐶𝑡𝑡=2 = 𝐹𝐹𝐶𝐶𝑡𝑡=1 + ∑ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=1𝑠𝑠𝑠𝑠 + ∑ (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=1𝑠𝑠𝑠𝑠
𝐹𝐹𝐶𝐶𝑡𝑡=3 = 𝐹𝐹𝐶𝐶𝑡𝑡=2 + ∑ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=2𝑠𝑠𝑠𝑠 + ∑ (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=2𝑠𝑠𝑠𝑠
∙∙∙
𝐹𝐹𝐶𝐶𝑡𝑡=𝑛𝑛 = 𝐹𝐹𝐶𝐶𝑡𝑡=𝑛𝑛−1 +∑ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=𝑛𝑛−1𝑠𝑠𝑠𝑠 +∑ (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡=𝑛𝑛−1𝑠𝑠
𝑠𝑠
𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠 = (𝐹𝐹𝐶𝐶𝑡𝑡 − ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 ) 𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠
∑ 𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠𝑠𝑠; 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠 = (𝐹𝐹𝐶𝐶𝑡𝑡 − ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 ) 𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠
∑ 𝑃𝑃𝑃𝑃𝑃𝑃𝑠𝑠𝑠𝑠
Given 0 ≤ ∅𝑠𝑠 ≤ 1, we have
𝑀𝑀𝑀𝑀𝑀𝑀{𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠,𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠} ≤ ∅𝑠𝑠 ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠 + (1 − ∅𝑠𝑠) ∙ 𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠 ≤ 𝑀𝑀𝑀𝑀𝑀𝑀{𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠,𝐶𝐶𝐶𝐶𝑃𝑃𝑡𝑡𝑠𝑠}.
This gives the sufficient condition for getting converged results at the end of the above process. Namely, when n →∞, 𝐹𝐹𝐶𝐶𝑡𝑡=𝑛𝑛 = ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 = ∑ 𝐶𝐶𝐶𝐶𝑃𝑃𝑠𝑠𝑠𝑠 .
97
Appendix D
WIOD country/region names WIOD sector classification
Code Country Code Name EU 15Annex Bused Code Description
C1 AUS Australia ✓ S1 Agriculture, Hunting, Forestry and FishingC2 AUT Austria ✓ ✓ S2* Mining and QuarryingC3 BEL Belgium ✓ ✓ S3 Food, Beverages and TobaccoC4 BGR Bulgaria ✓ S4 Textiles and Textile ProductsC5 BRA Brazil S5 Leather, Leather and FootwearC6 CAN Canada ✓ S6 Wood and Products of Wood and CorkC7 CHN China S7 Pulp, Paper, Paper , Printing and PublishingC8 CYP Cyprus S8* Coke, Refined Petroleum and Nuclear FuelC9 CZE Czech Republic ✓ S9 Chemicals and Chemical ProductsC10 DEU Germany ✓ ✓ S10 Rubber and PlasticsC11 DNK Denmark ✓ ✓ S11 Other Non-Metallic MineralC12 ESP Spain ✓ ✓ S12 Basic Metals and Fabricated MetalC13 EST Estonia ✓ S13 Machinery, NecC14 FIN Finland ✓ ✓ S14 Electrical and Optical EquipmentC15 FRA France ✓ ✓ S15 Transport EquipmentC16 GBR United Kingdom ✓ ✓ S16 Manufacturing, Nec; RecyclingC17 GRC Greece ✓ ✓ S17* Electricity, Gas and Water SupplyC18 HUN Hungary ✓ S18 ConstructionC19 IDN Indonesia S19 Sale, Maintenance and Repair of Motor Vehicles and Motorcycles; Retail Sale of FuelC20 IND India S20 Wholesale Trade and Commission Trade, Except of Motor Vehicles and MotorcyclesC21 IRL Ireland ✓ ✓ S21 Retail Trade, Except of Motor Vehicles and Motorcycles; Repair of Household GoodsC22 ITA Italy ✓ ✓ S22 Hotels and RestaurantsC23 JPN Japan ✓ S23 Inland TransportC24 KOR South Korea S24 Water TransportC25 LTU Lithuania ✓ S25 Air TransportC26 LUX Luxembourg ✓ ✓ S26 Other Supporting and Auxiliary Transport Activities; Activities of Travel AgenciesC27 LVA Latvia ✓ S27 Post and TelecommunicationsC28 MEX Mexico S28 Financial IntermediationC29 MLT Malta S29 Real Estate ActivitiesC30 NLD Netherlands ✓ ✓ S30 Renting of M&Eq and Other Business ActivitiesC31 POL Poland ✓ S31 Public Admin and Defence; Compulsory Social SecurityC32 PRT Portugal ✓ ✓ S32 EducationC33 ROM Romania ✓ S33 Health and Social WorkC34 RUS Russian Federation ✓ S34 Other Community, Social and Personal ServicesC35 SVK Slovakia ✓ S35 Private Households with Employed PersonsC36 SVN Slovenia ✓
C37 SWE Sweden ✓ ✓ *: energy related productsC38 TUR TurkeyC39 TWN TaiwanC40 USA United States ✓
C41 RoW Rest of the World